ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/OpenCL/OpenCL.xs
(Generate patch)

Comparing OpenCL/OpenCL.xs (file contents):
Revision 1.18 by root, Wed Nov 23 03:02:38 2011 UTC vs.
Revision 1.71 by root, Sat May 5 02:45:38 2012 UTC

1#include "EXTERN.h" 1#include "EXTERN.h"
2#include "perl.h" 2#include "perl.h"
3#include "XSUB.h" 3#include "XSUB.h"
4 4
5#include "ecb.h"//D
6
7#define X_STACKSIZE sizeof (void *) * 512 * 1024 // 2-4mb should be enough, really
8#include "xthread.h"
9#include "schmorp.h"
10
11#ifdef I_DLFCN
12 #include <dlfcn.h>
13#endif
14
15// how stupid is that, the 1.2 header files define CL_VERSION_1_1,
16// but then fail to define the api functions unless you ALSO define
17// this. This breaks 100% of the opencl 1.1 apps, for what reason?
18// after all, the functions are deprecated, not removed.
19// in addition, you cannot test for this in any future-proof way.
20// each time a new opencl version comes out, you need to make a new
21// release.
22#define CL_USE_DEPRECATED_OPENCL_1_2_APIS /* just guessing, you stupid idiots */
23
24#ifndef PREFER_1_1
25 #define PREFER_1_1 1
26#endif
27
28#if PREFER_1_1
29 #define CL_USE_DEPRECATED_OPENCL_1_1_APIS
30#endif
31
32#ifdef __APPLE__
33 #include <OpenCL/opencl.h>
34#else
5#include <CL/opencl.h> 35 #include <CL/opencl.h>
36#endif
37
38#ifndef CL_VERSION_1_2
39 #undef PREFER_1_1
40 #define PREFER_1_1 1
41#endif
6 42
7typedef cl_platform_id OpenCL__Platform; 43typedef cl_platform_id OpenCL__Platform;
8typedef cl_device_id OpenCL__Device; 44typedef cl_device_id OpenCL__Device;
45typedef cl_device_id OpenCL__SubDevice;
9typedef cl_context OpenCL__Context; 46typedef cl_context OpenCL__Context;
10typedef cl_command_queue OpenCL__Queue; 47typedef cl_command_queue OpenCL__Queue;
11typedef cl_mem OpenCL__Memory; 48typedef cl_mem OpenCL__Memory;
12typedef cl_mem OpenCL__Buffer; 49typedef cl_mem OpenCL__Buffer;
13typedef cl_mem OpenCL__BufferObj; 50typedef cl_mem OpenCL__BufferObj;
14typedef cl_mem OpenCL__Image; 51typedef cl_mem OpenCL__Image;
15typedef cl_mem OpenCL__Image2D;
16typedef cl_mem OpenCL__Image3D;
17typedef cl_mem OpenCL__Memory_ornull; 52typedef cl_mem OpenCL__Memory_ornull;
18typedef cl_mem OpenCL__Buffer_ornull; 53typedef cl_mem OpenCL__Buffer_ornull;
19typedef cl_mem OpenCL__Image_ornull; 54typedef cl_mem OpenCL__Image_ornull;
20typedef cl_mem OpenCL__Image2D_ornull;
21typedef cl_mem OpenCL__Image3D_ornull;
22typedef cl_sampler OpenCL__Sampler; 55typedef cl_sampler OpenCL__Sampler;
23typedef cl_program OpenCL__Program; 56typedef cl_program OpenCL__Program;
24typedef cl_kernel OpenCL__Kernel; 57typedef cl_kernel OpenCL__Kernel;
25typedef cl_event OpenCL__Event; 58typedef cl_event OpenCL__Event;
26typedef cl_event OpenCL__UserEvent; 59typedef cl_event OpenCL__UserEvent;
27 60
28typedef SV *FUTURE; 61typedef struct mapped * OpenCL__Mapped;
62
63static HV
64 *stash_platform,
65 *stash_device,
66 *stash_subdevice,
67 *stash_context,
68 *stash_queue,
69 *stash_program,
70 *stash_kernel,
71 *stash_sampler,
72 *stash_event,
73 *stash_userevent,
74 *stash_memory,
75 *stash_buffer,
76 *stash_bufferobj,
77 *stash_image,
78 *stash_image1d,
79 *stash_image1darray,
80 *stash_image1dbuffer,
81 *stash_image2d,
82 *stash_image2darray,
83 *stash_image3d,
84 *stash_mapped,
85 *stash_mappedbuffer,
86 *stash_mappedimage;
29 87
30/*****************************************************************************/ 88/*****************************************************************************/
31 89
90// name must include a leading underscore
91// all of this horrors would be unneceesary if somebody wrote a proper OpenGL module
92// for perl. doh.
93static void *
94glsym (const char *name)
95{
96 void *fun = 0;
97
98 #if defined I_DLFCN && defined RTLD_DEFAULT
99 fun = dlsym (RTLD_DEFAULT, name + 1);
100 if (!fun) fun = dlsym (RTLD_DEFAULT, name);
101
102 if (!fun)
103 {
104 static void *libgl;
105 static const char *glso[] = {
106 "libGL.so.1",
107 "libGL.so.3",
108 "libGL.so.4.0",
109 "libGL.so",
110 "/usr/lib/libGL.so",
111 "/usr/X11R6/lib/libGL.1.dylib"
112 };
113 int i;
114
115 for (i = 0; !libgl && i < sizeof (glso) / sizeof (glso [0]); ++i)
116 {
117 libgl = dlopen (glso [i], RTLD_LAZY);
118 if (libgl)
119 break;
120 }
121
122 if (libgl)
123 {
124 fun = dlsym (libgl, name + 1);
125 if (!fun) fun = dlsym (libgl, name);
126 }
127 }
128 #endif
129
130 return fun;
131}
132
133/*****************************************************************************/
134
32/* up to two temporary buffers */ 135/* up to two temporary buffers */
33static void * 136static void * ecb_noinline
34tmpbuf (size_t size) 137tmpbuf (size_t size)
35{ 138{
139 enum { buffers = 4 };
36 static int idx; 140 static int idx;
37 static void *buf [2]; 141 static void *buf [buffers];
38 static size_t len [2]; 142 static size_t len [buffers];
39 143
40 idx ^= 1; 144 idx = (idx + 1) % buffers;
41 145
42 if (len [idx] < size) 146 if (len [idx] < size)
43 { 147 {
44 free (buf [idx]); 148 free (buf [idx]);
45 len [idx] = ((size + 31) & ~4095) + 4096 - 32; 149 len [idx] = ((size + 31) & ~4095) + 4096 - 32;
46 buf [idx] = malloc (len [idx]); 150 buf [idx] = malloc (len [idx]);
47 } 151 }
48 152
49 return buf [idx]; 153 return buf [idx];
154}
155
156static const char * ecb_noinline
157cv_get_name (CV *cv)
158{
159 static char fullname [256];
160
161 GV *gv = CvGV (cv); // gv better be nonzero
162
163 HV *stash = GvSTASH (gv);
164 const char *hvname = HvNAME_get (stash); // stash also better be nonzero
165 const char *gvname = GvNAME (gv);
166
167 snprintf (fullname, sizeof (fullname), "%s::%s", hvname, gvname);
168 return fullname;
50} 169}
51 170
52/*****************************************************************************/ 171/*****************************************************************************/
53 172
54typedef struct 173typedef struct
113 if (res) \ 232 if (res) \
114 FAIL (name); 233 FAIL (name);
115 234
116/*****************************************************************************/ 235/*****************************************************************************/
117 236
118#define NEW_MORTAL_OBJ(class,ptr) sv_setref_pv (sv_newmortal (), class, ptr)
119#define XPUSH_NEW_OBJ(class,ptr) XPUSHs (NEW_MORTAL_OBJ (class, ptr))
120
121static void * 237static SV *
238new_clobj (HV *stash, IV id)
239{
240 return sv_2mortal (sv_bless (newRV_noinc (newSViv (id)), stash));
241}
242
243#define PUSH_CLOBJ(stash,id) PUSHs (new_clobj ((stash), (IV)(id)))
244#define XPUSH_CLOBJ(stash,id) XPUSHs (new_clobj ((stash), (IV)(id)))
245
246/* cl objects are either \$iv, or [$iv, ...] */
247/* they can be upgraded at runtime to the array form */
248static void * ecb_noinline
122SvPTROBJ (const char *func, const char *svname, SV *sv, const char *pkg) 249SvCLOBJ (CV *cv, const char *svname, SV *sv, const char *pkg)
123{ 250{
251 // sv_derived_from is quite slow :(
124 if (SvROK (sv) && sv_derived_from (sv, pkg)) 252 if (SvROK (sv) && sv_derived_from (sv, pkg))
125 return (void *)SvIV (SvRV (sv)); 253 return (void *)SvIV (SvRV (sv));
126 254
127 croak ("%s: %s is not of type %s", func, svname, pkg); 255 croak ("%s: %s is not of type %s", cv_get_name (cv), svname, pkg);
256}
257
258// the "no-inherit" version of the above
259static void * ecb_noinline
260SvCLOBJ_ni (CV *cv, const char *svname, SV *sv, HV *stash)
261{
262 if (SvROK (sv) && SvSTASH (SvRV (sv)) == stash)
263 return (void *)SvIV (SvRV (sv));
264
265 croak ("%s: %s is not of type %s", cv_get_name (cv), svname, HvNAME (stash));
128} 266}
129 267
130/*****************************************************************************/ 268/*****************************************************************************/
269
270static cl_context_properties * ecb_noinline
271SvCONTEXTPROPERTIES (CV *cv, const char *svname, SV *sv, cl_context_properties *extra, int extracount)
272{
273 if (!sv || !SvOK (sv))
274 if (extra)
275 sv = sv_2mortal (newRV_noinc ((SV *)newAV ())); // slow, but rarely used hopefully
276 else
277 return 0;
278
279 if (SvROK (sv) && SvTYPE (SvRV (sv)) == SVt_PVAV)
280 {
281 AV *av = (AV *)SvRV (sv);
282 int i, len = av_len (av) + 1;
283 cl_context_properties *p = tmpbuf (sizeof (cl_context_properties) * (len + extracount + 1));
284 cl_context_properties *l = p;
285
286 if (len & 1)
287 croak ("%s: %s is not a property list (must contain an even number of elements)", cv_get_name (cv), svname);
288
289 while (extracount--)
290 *l++ = *extra++;
291
292 for (i = 0; i < len; i += 2)
293 {
294 cl_context_properties t = SvIV (*av_fetch (av, i , 0));
295 SV *p_sv = *av_fetch (av, i + 1, 0);
296 cl_context_properties v = SvIV (p_sv); // code below can override
297
298 switch (t)
299 {
300 case CL_CONTEXT_PLATFORM:
301 if (SvROK (p_sv))
302 v = (cl_context_properties)SvCLOBJ (cv, svname, p_sv, "OpenCL::Platform");
303 break;
304
305 case CL_GLX_DISPLAY_KHR:
306 if (!SvOK (p_sv))
307 {
308 void *func = glsym ("_glXGetCurrentDisplay");
309 if (func)
310 v = (cl_context_properties)((void *(*)(void))func)();
311 }
312 break;
313
314 case CL_GL_CONTEXT_KHR:
315 if (!SvOK (p_sv))
316 {
317 void *func = glsym ("_glXGetCurrentContext");
318 if (func)
319 v = (cl_context_properties)((void *(*)(void))func)();
320 }
321 break;
322
323 default:
324 /* unknown property, treat as int */
325 break;
326 }
327
328 *l++ = t;
329 *l++ = v;
330 }
331
332 *l = 0;
333
334 return p;
335 }
336
337 croak ("%s: %s is not a property list (either undef or [type => value, ...])", cv_get_name (cv), svname);
338}
339
340// parse an array of CLOBJ into a void ** array in C - works only for CLOBJs whose representation
341// is a pointer (and only on well-behaved systems).
342static void * ecb_noinline
343object_list (CV *cv, int or_undef, const char *argname, SV *arg, const char *klass, cl_uint *rcount)
344{
345 if (!SvROK (arg) || SvTYPE (SvRV (arg)) != SVt_PVAV)
346 croak ("%s: '%s' parameter must be %sa reference to an array of %s objects",
347 cv_get_name (cv), argname, or_undef ? "undef or " : "", klass);
348
349 AV *av = (AV *)SvRV (arg);
350 void **list = 0;
351 cl_uint count = av_len (av) + 1;
352
353 if (count)
354 {
355 list = tmpbuf (sizeof (*list) * count);
356 int i;
357 for (i = 0; i < count; ++i)
358 list [i] = SvCLOBJ (cv, argname, *av_fetch (av, i, 1), klass);
359 }
360
361 if (!count && !or_undef)
362 croak ("%s: '%s' must contain at least one %s object",
363 cv_get_name (cv), argname, klass);
364
365 *rcount = count;
366 return (void *)list;
367}
368
369/*****************************************************************************/
370/* callback stuff */
371
372/* default context callback, log to stderr */
373static void CL_CALLBACK
374context_default_notify (const char *msg, const void *info, size_t cb, void *data)
375{
376 fprintf (stderr, "OpenCL Context Notify: %s\n", msg);
377}
378
379typedef struct
380{
381 int free_cb;
382 void (*push)(void *data1, void *data2, void *data3);
383} eq_vtbl;
384
385typedef struct eq_item
386{
387 struct eq_item *next;
388 eq_vtbl *vtbl;
389 SV *cb;
390 void *data1, *data2, *data3;
391} eq_item;
392
393static void (*eq_signal_func)(void *signal_arg, int value);
394static void *eq_signal_arg;
395static xmutex_t eq_lock = X_MUTEX_INIT;
396static eq_item *eq_head, *eq_tail;
397
398static void ecb_noinline
399eq_enq (eq_vtbl *vtbl, SV *cb, void *data1, void *data2, void *data3)
400{
401 eq_item *item = malloc (sizeof (eq_item));
402
403 item->next = 0;
404 item->vtbl = vtbl;
405 item->cb = cb;
406 item->data1 = data1;
407 item->data2 = data2;
408 item->data3 = data3;
409
410 X_LOCK (eq_lock);
411
412 *(eq_head ? &eq_tail->next : &eq_head) = item;
413 eq_tail = item;
414
415 X_UNLOCK (eq_lock);
416
417 eq_signal_func (eq_signal_arg, 0);
418}
419
420static eq_item *
421eq_dec (void)
422{
423 eq_item *res;
424
425 X_LOCK (eq_lock);
426
427 res = eq_head;
428 if (res)
429 eq_head = res->next;
430
431 X_UNLOCK (eq_lock);
432
433 return res;
434}
435
436static void
437eq_poll (void)
438{
439 eq_item *item;
440
441 while ((item = eq_dec ()))
442 {
443 ENTER;
444 SAVETMPS;
445
446 dSP;
447 PUSHMARK (SP);
448 EXTEND (SP, 2);
449
450 if (item->vtbl->free_cb)
451 sv_2mortal (item->cb);
452
453 PUTBACK;
454 item->vtbl->push (item->data1, item->data2, item->data3);
455
456 SV *cb = item->cb;
457 free (item);
458
459 call_sv (cb, G_DISCARD | G_VOID);
460
461 FREETMPS;
462 LEAVE;
463 }
464}
465
466static void
467eq_poll_interrupt (pTHX_ void *c_arg, int value)
468{
469 eq_poll ();
470}
471
472/*****************************************************************************/
473/* context notify */
474
475static void ecb_noinline
476eq_context_push (void *data1, void *data2, void *data3)
477{
478 dSP;
479 PUSHs (sv_2mortal (newSVpv (data1, 0)));
480 PUSHs (sv_2mortal (newSVpvn (data2, (STRLEN)data3)));
481 PUTBACK;
482
483 free (data1);
484 free (data2);
485}
486
487static eq_vtbl eq_context_vtbl = { 0, eq_context_push };
488
489static void CL_CALLBACK
490eq_context_notify (const char *msg, const void *pvt, size_t cb, void *user_data)
491{
492 void *pvt_copy = malloc (cb);
493 memcpy (pvt_copy, pvt, cb);
494 eq_enq (&eq_context_vtbl, user_data, strdup (msg), pvt_copy, (void *)cb);
495}
496
497#define CONTEXT_NOTIFY_CALLBACK \
498 void (CL_CALLBACK *pfn_notify)(const char *, const void *, size_t, void *) = context_default_notify; \
499 void *user_data = 0; \
500 \
501 if (SvOK (notify)) \
502 { \
503 pfn_notify = eq_context_notify; \
504 user_data = s_get_cv (notify); \
505 }
506
507static SV * ecb_noinline
508new_clobj_context (cl_context ctx, void *user_data)
509{
510 SV *sv = new_clobj (stash_context, (IV)ctx);
511
512 if (user_data)
513 sv_magicext (SvRV (sv), user_data, PERL_MAGIC_ext, 0, 0, 0);
514
515 return sv;
516}
517
518#define XPUSH_CLOBJ_CONTEXT XPUSHs (new_clobj_context (ctx, user_data));
519
520/*****************************************************************************/
521/* build/compile/link notify */
522
523static void
524eq_program_push (void *data1, void *data2, void *data3)
525{
526 dSP;
527 PUSH_CLOBJ (stash_program, data1);
528 PUTBACK;
529}
530
531static eq_vtbl eq_program_vtbl = { 1, eq_program_push };
532
533static void CL_CALLBACK
534eq_program_notify (cl_program program, void *user_data)
535{
536 clRetainProgram (program);
537
538 eq_enq (&eq_program_vtbl, user_data, (void *)program, 0, 0);
539}
540
541typedef void (CL_CALLBACK *program_callback)(cl_program program, void *user_data);
542
543static program_callback ecb_noinline
544make_program_callback (SV *notify, void **ruser_data)
545{
546 if (SvOK (notify))
547 {
548 *ruser_data = SvREFCNT_inc (s_get_cv (notify));
549 return eq_program_notify;
550 }
551 else
552 {
553 *ruser_data = 0;
554 return 0;
555 }
556}
557
558struct build_args
559{
560 cl_program program;
561 char *options;
562 void *user_data;
563 cl_uint num_devices;
564};
565
566X_THREAD_PROC (build_program_thread)
567{
568 struct build_args *arg = thr_arg;
569
570 clBuildProgram (arg->program, arg->num_devices, arg->num_devices ? (void *)(arg + 1) : 0, arg->options, 0, 0);
571
572 if (arg->user_data)
573 eq_program_notify (arg->program, arg->user_data);
574 else
575 clReleaseProgram (arg->program);
576
577 free (arg->options);
578 free (arg);
579
580 return 0;
581}
582
583static void
584build_program_async (cl_program program, cl_uint num_devices, const cl_device_id *device_list, const char *options, void *user_data)
585{
586 struct build_args *arg = malloc (sizeof (struct build_args) + sizeof (*device_list) * num_devices);
587
588 arg->program = program;
589 arg->options = strdup (options);
590 arg->user_data = user_data;
591 arg->num_devices = num_devices;
592 memcpy (arg + 1, device_list, sizeof (*device_list) * num_devices);
593
594 xthread_t id;
595 thread_create (&id, build_program_thread, arg);
596}
597
598/*****************************************************************************/
599/* event objects */
600
601static void
602eq_event_push (void *data1, void *data2, void *data3)
603{
604 dSP;
605 PUSH_CLOBJ (stash_event, data1);
606 PUSHs (sv_2mortal (newSViv ((IV)data2)));
607 PUTBACK;
608}
609
610static eq_vtbl eq_event_vtbl = { 1, eq_event_push };
611
612static void CL_CALLBACK
613eq_event_notify (cl_event event, cl_int event_command_exec_status, void *user_data)
614{
615 clRetainEvent (event);
616 eq_enq (&eq_event_vtbl, user_data, (void *)event, (void *)(IV)event_command_exec_status, 0);
617}
618
619/*****************************************************************************/
620/* utilities for XS code */
131 621
132static size_t 622static size_t
133img_row_pitch (cl_mem img) 623img_row_pitch (cl_mem img)
134{ 624{
135 size_t res; 625 size_t res;
136 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0); 626 clGetImageInfo (img, CL_IMAGE_ROW_PITCH, sizeof (res), &res, 0);
137 return res; 627 return res;
138} 628}
139 629
140static cl_event * 630static cl_event * ecb_noinline
141event_list (SV **items, int count) 631event_list (CV *cv, SV **items, cl_uint *rcount, cl_event extra)
142{ 632{
633 cl_uint count = *rcount;
634
635 if (count > 0x7fffffffU) // yeah, it's a hack - the caller might have underflowed
636 *rcount = count = 0;
637
638 if (!count && !extra)
639 return 0;
640
143 cl_event *list = tmpbuf (sizeof (cl_event) * count); 641 cl_event *list = tmpbuf (sizeof (cl_event) * (count + 1));
642 int i = 0;
144 643
145 while (count--) 644 while (count--)
645 if (SvOK (items [count]))
146 list [count] = SvPTROBJ ("clEnqueue", "wait_events", items [count], "OpenCL::Event"); 646 list [i++] = SvCLOBJ (cv, "wait_events", items [count], "OpenCL::Event");
147 647
648 if (extra)
649 list [i++] = extra;
650
651 *rcount = i;
652
148 return list; 653 return i ? list : 0;
149} 654}
150 655
151#define EVENT_LIST(items,count) \ 656#define EVENT_LIST(skip) \
152 cl_uint event_list_count = (count); \ 657 cl_uint event_list_count = items - (skip); \
153 cl_event *event_list_ptr = event_list (&ST (items), event_list_count) 658 cl_event *event_list_ptr = event_list (cv, &ST (skip), &event_list_count, 0)
154 659
155#define INFO(class) \ 660#define INFO(class) \
156{ \ 661{ \
157 size_t size; \ 662 size_t size; \
158 NEED_SUCCESS (Get ## class ## Info, (this, name, 0, 0, &size)); \ 663 NEED_SUCCESS (Get ## class ## Info, (self, name, 0, 0, &size)); \
159 SV *sv = sv_2mortal (newSV (size)); \ 664 SV *sv = sv_2mortal (newSV (size)); \
160 SvUPGRADE (sv, SVt_PV); \ 665 SvUPGRADE (sv, SVt_PV); \
161 SvPOK_only (sv); \ 666 SvPOK_only (sv); \
162 SvCUR_set (sv, size); \ 667 SvCUR_set (sv, size); \
163 NEED_SUCCESS (Get ## class ## Info, (this, name, size, SvPVX (sv), 0)); \ 668 NEED_SUCCESS (Get ## class ## Info, (self, name, size, SvPVX (sv), 0)); \
164 XPUSHs (sv); \ 669 XPUSHs (sv); \
165} 670}
166 671
672/*****************************************************************************/
673/* mapped_xxx */
674
675static OpenCL__Mapped
676SvMAPPED (SV *sv)
677{
678 // no typechecking atm., keep your fingers crossed
679 return (OpenCL__Mapped)SvMAGIC (SvRV (sv))->mg_ptr;
680}
681
682struct mapped
683{
684 cl_command_queue queue;
685 cl_mem memobj;
686 void *ptr;
687 size_t cb;
688 cl_event event;
689 size_t row_pitch;
690 size_t slice_pitch;
691};
692
693static SV *
694mapped_new (HV *stash, cl_command_queue queue, cl_mem memobj, cl_map_flags flags, void *ptr, size_t cb, cl_event ev, size_t row_pitch, size_t slice_pitch)
695{
696 SV *data = newSV (0);
697 SvUPGRADE (data, SVt_PVMG);
698
699 OpenCL__Mapped mapped;
700 New (0, mapped, 1, struct mapped);
701
702 clRetainCommandQueue (queue);
703
704 mapped->queue = queue;
705 mapped->memobj = memobj;
706 mapped->ptr = ptr;
707 mapped->cb = cb;
708 mapped->event = ev;
709 mapped->row_pitch = row_pitch;
710 mapped->slice_pitch = slice_pitch;
711
712 sv_magicext (data, 0, PERL_MAGIC_ext, 0, (char *)mapped, 0);
713
714 if (SvLEN (data))
715 Safefree (data);
716
717 SvPVX (data) = (char *)ptr;
718 SvCUR_set (data, cb);
719 SvLEN_set (data, 0);
720 SvPOK_only (data);
721
722 SV *obj = sv_2mortal (sv_bless (newRV_noinc (data), stash));
723
724 if (!(flags & CL_MAP_WRITE))
725 SvREADONLY_on (data);
726
727 return obj;
728}
729
730static void
731mapped_detach (SV *sv, OpenCL__Mapped mapped)
732{
733 SV *data = SvRV (sv);
734
735 // the next check checks both before AND after detach, where SvPVX should STILL be 0
736 if (SvPVX (data) != (char *)mapped->ptr)
737 warn ("FATAL: OpenCL memory mapped scalar changed location, detected");
738 else
739 {
740 SvREADONLY_off (data);
741 SvCUR_set (data, 0);
742 SvPVX (data) = 0;
743 SvOK_off (data);
744 }
745
746 mapped->ptr = 0;
747}
748
749static void
750mapped_unmap (CV *cv, SV *self, OpenCL__Mapped mapped, cl_command_queue queue, SV **wait_list, cl_uint event_list_count)
751{
752 cl_event *event_list_ptr = event_list (cv, wait_list, &event_list_count, mapped->event);
753 cl_event ev;
754
755 NEED_SUCCESS (EnqueueUnmapMemObject, (queue, mapped->memobj, mapped->ptr, event_list_count, event_list_ptr, &ev));
756
757 clReleaseEvent (mapped->event);
758 mapped->event = ev;
759
760 mapped_detach (self, mapped);
761}
762
763/*****************************************************************************/
764
167MODULE = OpenCL PACKAGE = OpenCL 765MODULE = OpenCL PACKAGE = OpenCL
168 766
169PROTOTYPES: ENABLE 767PROTOTYPES: ENABLE
170 768
769void
770poll ()
771 CODE:
772 eq_poll ();
773
774void
775_eq_initialise (IV func, IV arg)
776 CODE:
777 eq_signal_func = (void (*)(void *, int))func;
778 eq_signal_arg = (void*)arg;
779
171BOOT: 780BOOT:
172{ 781{
173 HV *stash = gv_stashpv ("OpenCL", 1); 782 HV *stash = gv_stashpv ("OpenCL", 1);
174 static const ivstr *civ, const_iv[] = { 783 static const ivstr *civ, const_iv[] = {
175 { sizeof (cl_char ), "SIZEOF_CHAR" }, 784 { sizeof (cl_char ), "SIZEOF_CHAR" },
176 { sizeof (cl_uchar ), "SIZEOF_UCHAR" }, 785 { sizeof (cl_uchar ), "SIZEOF_UCHAR" },
177 { sizeof (cl_short ), "SIZEOF_SHORT" }, 786 { sizeof (cl_short ), "SIZEOF_SHORT" },
178 { sizeof (cl_ushort), "SIZEOF_USHORT" }, 787 { sizeof (cl_ushort), "SIZEOF_USHORT" },
179 { sizeof (cl_int ), "SIZEOF_INT" }, 788 { sizeof (cl_int ), "SIZEOF_INT" },
180 { sizeof (cl_uint ), "SIZEOF_UINT" }, 789 { sizeof (cl_uint ), "SIZEOF_UINT" },
181 { sizeof (cl_long ), "SIZEOF_LONG" }, 790 { sizeof (cl_long ), "SIZEOF_LONG" },
182 { sizeof (cl_ulong ), "SIZEOF_ULONG" }, 791 { sizeof (cl_ulong ), "SIZEOF_ULONG" },
183 { sizeof (cl_half ), "SIZEOF_HALF" }, 792 { sizeof (cl_half ), "SIZEOF_HALF" },
184 { sizeof (cl_float ), "SIZEOF_FLOAT" }, 793 { sizeof (cl_float ), "SIZEOF_FLOAT" },
185 { sizeof (cl_double), "SIZEOF_DOUBLE" }, 794 { sizeof (cl_double), "SIZEOF_DOUBLE" },
186#include "constiv.h" 795#include "constiv.h"
187 }; 796 };
797
188 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--) 798 for (civ = const_iv + sizeof (const_iv) / sizeof (const_iv [0]); civ > const_iv; civ--)
189 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv)); 799 newCONSTSUB (stash, (char *)civ[-1].name, newSViv (civ[-1].iv));
800
801 stash_platform = gv_stashpv ("OpenCL::Platform", GV_ADD);
802 stash_device = gv_stashpv ("OpenCL::Device", GV_ADD);
803 stash_subdevice = gv_stashpv ("OpenCL::SubDevice", GV_ADD);
804 stash_context = gv_stashpv ("OpenCL::Context", GV_ADD);
805 stash_queue = gv_stashpv ("OpenCL::Queue", GV_ADD);
806 stash_program = gv_stashpv ("OpenCL::Program", GV_ADD);
807 stash_kernel = gv_stashpv ("OpenCL::Kernel", GV_ADD);
808 stash_sampler = gv_stashpv ("OpenCL::Sampler", GV_ADD);
809 stash_event = gv_stashpv ("OpenCL::Event", GV_ADD);
810 stash_userevent = gv_stashpv ("OpenCL::UserEvent", GV_ADD);
811 stash_memory = gv_stashpv ("OpenCL::Memory", GV_ADD);
812 stash_buffer = gv_stashpv ("OpenCL::Buffer", GV_ADD);
813 stash_bufferobj = gv_stashpv ("OpenCL::BufferObj", GV_ADD);
814 stash_image = gv_stashpv ("OpenCL::Image", GV_ADD);
815 stash_image1d = gv_stashpv ("OpenCL::Image1D", GV_ADD);
816 stash_image1darray = gv_stashpv ("OpenCL::Image1DArray", GV_ADD);
817 stash_image1dbuffer = gv_stashpv ("OpenCL::Image1DBuffer", GV_ADD);
818 stash_image2d = gv_stashpv ("OpenCL::Image2D", GV_ADD);
819 stash_image2darray = gv_stashpv ("OpenCL::Image2DArray", GV_ADD);
820 stash_image3d = gv_stashpv ("OpenCL::Image3D", GV_ADD);
821 stash_mapped = gv_stashpv ("OpenCL::Mapped", GV_ADD);
822 stash_mappedbuffer = gv_stashpv ("OpenCL::MappedBuffer", GV_ADD);
823 stash_mappedimage = gv_stashpv ("OpenCL::MappedImage", GV_ADD);
824
825 sv_setiv (perl_get_sv ("OpenCL::POLL_FUNC", TRUE), (IV)eq_poll_interrupt);
190} 826}
191 827
192cl_int 828cl_int
193errno () 829errno ()
194 CODE: 830 CODE:
195 errno = res; 831 RETVAL = res;
832 OUTPUT:
833 RETVAL
196 834
197const char * 835const char *
198err2str (cl_int err) 836err2str (cl_int err = res)
199 837
200const char * 838const char *
201enum2str (cl_uint value) 839enum2str (cl_uint value)
202 840
203void 841void
211 list = tmpbuf (sizeof (*list) * count); 849 list = tmpbuf (sizeof (*list) * count);
212 NEED_SUCCESS (GetPlatformIDs, (count, list, 0)); 850 NEED_SUCCESS (GetPlatformIDs, (count, list, 0));
213 851
214 EXTEND (SP, count); 852 EXTEND (SP, count);
215 for (i = 0; i < count; ++i) 853 for (i = 0; i < count; ++i)
216 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Platform", list [i])); 854 PUSH_CLOBJ (stash_platform, list [i]);
217 855
218void 856void
219context_from_type (FUTURE properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, FUTURE notify = 0) 857context_from_type (cl_context_properties *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
220 PPCODE: 858 PPCODE:
859 CONTEXT_NOTIFY_CALLBACK;
221 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (0, type, 0, 0, &res)); 860 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (properties, type, pfn_notify, user_data, &res));
222 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 861 XPUSH_CLOBJ_CONTEXT;
223 862
224void 863void
225context (FUTURE properties, FUTURE devices, FUTURE notify = 0) 864context (cl_context_properties *properties, SV *devices, SV *notify = &PL_sv_undef)
226 PPCODE: 865 PPCODE:
227 /* der Gipfel der Kunst */ 866 cl_uint device_count;
867 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
868
869 CONTEXT_NOTIFY_CALLBACK;
870 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (properties, device_count, device_list, pfn_notify, user_data, &res));
871 XPUSH_CLOBJ_CONTEXT;
228 872
229void 873void
230wait_for_events (...) 874wait_for_events (...)
231 CODE: 875 CODE:
232 EVENT_LIST (0, items); 876 EVENT_LIST (0);
233 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr)); 877 NEED_SUCCESS (WaitForEvents, (event_list_count, event_list_ptr));
234 878
235PROTOTYPES: DISABLE 879PROTOTYPES: DISABLE
236 880
237MODULE = OpenCL PACKAGE = OpenCL::Platform 881MODULE = OpenCL PACKAGE = OpenCL::Platform
238 882
239void 883void
240info (OpenCL::Platform this, cl_platform_info name) 884info (OpenCL::Platform self, cl_platform_info name)
241 PPCODE: 885 PPCODE:
242 INFO (Platform) 886 INFO (Platform)
243 887
888void
889unload_compiler (OpenCL::Platform self)
890 CODE:
891#if CL_VERSION_1_2
892 clUnloadPlatformCompiler (self);
893#endif
894
244#BEGIN:platform 895#BEGIN:platform
245 896
246void 897void
247profile (OpenCL::Platform this) 898profile (OpenCL::Platform self)
248 ALIAS: 899 ALIAS:
249 profile = CL_PLATFORM_PROFILE 900 profile = CL_PLATFORM_PROFILE
250 version = CL_PLATFORM_VERSION 901 version = CL_PLATFORM_VERSION
251 name = CL_PLATFORM_NAME 902 name = CL_PLATFORM_NAME
252 vendor = CL_PLATFORM_VENDOR 903 vendor = CL_PLATFORM_VENDOR
253 extensions = CL_PLATFORM_EXTENSIONS 904 extensions = CL_PLATFORM_EXTENSIONS
254 PPCODE: 905 PPCODE:
255 size_t size; 906 size_t size;
256 NEED_SUCCESS (GetPlatformInfo, (this, ix, 0, 0, &size)); 907 NEED_SUCCESS (GetPlatformInfo, (self, ix, 0, 0, &size));
257 char *value = tmpbuf (size); 908 char *value = tmpbuf (size);
258 NEED_SUCCESS (GetPlatformInfo, (this, ix, size, value, 0)); 909 NEED_SUCCESS (GetPlatformInfo, (self, ix, size, value, 0));
259 EXTEND (SP, 1); 910 EXTEND (SP, 1);
260 const int i = 0; 911 const int i = 0;
261 PUSHs (sv_2mortal (newSVpv (value, 0))); 912 PUSHs (sv_2mortal (newSVpv (value, 0)));
262 913
263#END:platform 914#END:platform
264 915
265void 916void
266devices (OpenCL::Platform this, cl_device_type type = CL_DEVICE_TYPE_ALL) 917devices (OpenCL::Platform self, cl_device_type type = CL_DEVICE_TYPE_ALL)
267 PPCODE: 918 PPCODE:
268 cl_device_id *list; 919 cl_device_id *list;
269 cl_uint count; 920 cl_uint count;
270 int i; 921 int i;
271 922
272 NEED_SUCCESS (GetDeviceIDs, (this, type, 0, 0, &count)); 923 NEED_SUCCESS (GetDeviceIDs, (self, type, 0, 0, &count));
273 list = tmpbuf (sizeof (*list) * count); 924 list = tmpbuf (sizeof (*list) * count);
274 NEED_SUCCESS (GetDeviceIDs, (this, type, count, list, 0)); 925 NEED_SUCCESS (GetDeviceIDs, (self, type, count, list, 0));
275 926
276 EXTEND (SP, count); 927 EXTEND (SP, count);
277 for (i = 0; i < count; ++i) 928 for (i = 0; i < count; ++i)
278 PUSHs (sv_setref_pv (sv_newmortal (), "OpenCL::Device", list [i])); 929 PUSH_CLOBJ (stash_device, list [i]);
279 930
280void 931void
281context (OpenCL::Platform this, FUTURE properties, SV *devices, FUTURE notify = 0) 932context (OpenCL::Platform self, SV *properties, SV *devices, SV *notify = &PL_sv_undef)
282 PPCODE: 933 PPCODE:
934 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
935 cl_context_properties *props = SvCONTEXTPROPERTIES (cv, "properties", properties, extra, 2);
936
937 cl_uint device_count;
938 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
939
940 CONTEXT_NOTIFY_CALLBACK;
941 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (props, device_count, device_list, pfn_notify, user_data, &res));
942 XPUSH_CLOBJ_CONTEXT;
943
944void
945context_from_type (OpenCL::Platform self, SV *properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, SV *notify = &PL_sv_undef)
946 PPCODE:
947 cl_context_properties extra[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)self };
948 cl_context_properties *props = SvCONTEXTPROPERTIES (cv, "properties", properties, extra, 2);
949
950 CONTEXT_NOTIFY_CALLBACK;
951 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, pfn_notify, user_data, &res));
952 XPUSH_CLOBJ_CONTEXT;
953
954MODULE = OpenCL PACKAGE = OpenCL::Device
955
956void
957info (OpenCL::Device self, cl_device_info name)
958 PPCODE:
959 INFO (Device)
960
961#if CL_VERSION_1_2
962
963void
964sub_devices (OpenCL::Device self, SV *properties)
965 PPCODE:
283 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVAV) 966 if (!SvROK (properties) || SvTYPE (SvRV (properties)) != SVt_PVAV)
284 croak ("OpenCL::Platform argument 'device' must be an arrayref with device objects, in call"); 967 croak ("OpenCL::Device::sub_devices: properties must be specified as reference to an array of property-value pairs");
285 968
286 AV *av = (AV *)SvRV (devices); 969 properties = SvRV (properties);
287 cl_uint num_devices = av_len (av) + 1; 970
288 cl_device_id *device_list = tmpbuf (sizeof (cl_device_id) * num_devices); 971 cl_uint count = av_len ((AV *)properties) + 1;
972 cl_device_partition_property *props = tmpbuf (sizeof (*props) * count + 1);
973
289 int i; 974 int i;
975 for (i = 0; i < count; ++i)
976 props [i] = (cl_device_partition_property)SvIV (*av_fetch ((AV *)properties, i, 0));
290 977
291 for (i = num_devices; i--; ) 978 props [count] = 0;
292 device_list [i] = SvPTROBJ ("clCreateContext", "devices", *av_fetch (av, i, 0), "OpenCL::Device");
293 979
294 NEED_SUCCESS_ARG (cl_context ctx, CreateContext, (0, num_devices, device_list, 0, 0, &res)); 980 cl_uint num_devices;
295 XPUSH_NEW_OBJ ("OpenCL::Context", ctx); 981 NEED_SUCCESS (CreateSubDevices, (self, props, 0, 0, &num_devices));
982 cl_device_id *list = tmpbuf (sizeof (*list) * num_devices);
983 NEED_SUCCESS (CreateSubDevices, (self, props, num_devices, list, 0));
296 984
297void 985 EXTEND (SP, num_devices);
298context_from_type (OpenCL::Platform this, FUTURE properties = 0, cl_device_type type = CL_DEVICE_TYPE_DEFAULT, FUTURE notify = 0) 986 for (i = 0; i < count; ++i)
299 PPCODE: 987 PUSH_CLOBJ (stash_subdevice, list [i]);
300 cl_context_properties props[] = { CL_CONTEXT_PLATFORM, (cl_context_properties)this, 0 };
301 NEED_SUCCESS_ARG (cl_context ctx, CreateContextFromType, (props, type, 0, 0, &res));
302 XPUSH_NEW_OBJ ("OpenCL::Context", ctx);
303 988
304MODULE = OpenCL PACKAGE = OpenCL::Device 989#endif
305
306void
307info (OpenCL::Device this, cl_device_info name)
308 PPCODE:
309 INFO (Device)
310 990
311#BEGIN:device 991#BEGIN:device
312 992
313void 993void
314type (OpenCL::Device this) 994type (OpenCL::Device self)
315 PPCODE: 995 PPCODE:
316 cl_device_type value [1]; 996 cl_device_type value [1];
317 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_TYPE, sizeof (value), value, 0)); 997 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_TYPE, sizeof (value), value, 0));
318 EXTEND (SP, 1); 998 EXTEND (SP, 1);
319 const int i = 0; 999 const int i = 0;
320 PUSHs (sv_2mortal (newSViv (value [i]))); 1000 PUSHs (sv_2mortal (newSVuv (value [i])));
321 1001
322void 1002void
323vendor_id (OpenCL::Device this) 1003vendor_id (OpenCL::Device self)
324 ALIAS: 1004 ALIAS:
325 vendor_id = CL_DEVICE_VENDOR_ID 1005 vendor_id = CL_DEVICE_VENDOR_ID
326 max_compute_units = CL_DEVICE_MAX_COMPUTE_UNITS 1006 max_compute_units = CL_DEVICE_MAX_COMPUTE_UNITS
327 max_work_item_dimensions = CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS 1007 max_work_item_dimensions = CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS
328 preferred_vector_width_char = CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR 1008 preferred_vector_width_char = CL_DEVICE_PREFERRED_VECTOR_WIDTH_CHAR
346 native_vector_width_int = CL_DEVICE_NATIVE_VECTOR_WIDTH_INT 1026 native_vector_width_int = CL_DEVICE_NATIVE_VECTOR_WIDTH_INT
347 native_vector_width_long = CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG 1027 native_vector_width_long = CL_DEVICE_NATIVE_VECTOR_WIDTH_LONG
348 native_vector_width_float = CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT 1028 native_vector_width_float = CL_DEVICE_NATIVE_VECTOR_WIDTH_FLOAT
349 native_vector_width_double = CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE 1029 native_vector_width_double = CL_DEVICE_NATIVE_VECTOR_WIDTH_DOUBLE
350 native_vector_width_half = CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF 1030 native_vector_width_half = CL_DEVICE_NATIVE_VECTOR_WIDTH_HALF
351 reference_count_ext = CL_DEVICE_REFERENCE_COUNT_EXT 1031 reference_count_ext = CL_DEVICE_REFERENCE_COUNT_EXT
352 PPCODE: 1032 PPCODE:
353 cl_uint value [1]; 1033 cl_uint value [1];
354 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1034 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
355 EXTEND (SP, 1); 1035 EXTEND (SP, 1);
356 const int i = 0; 1036 const int i = 0;
357 PUSHs (sv_2mortal (newSVuv (value [i]))); 1037 PUSHs (sv_2mortal (newSVuv (value [i])));
358 1038
359void 1039void
360max_work_group_size (OpenCL::Device this) 1040max_work_group_size (OpenCL::Device self)
361 ALIAS: 1041 ALIAS:
362 max_work_group_size = CL_DEVICE_MAX_WORK_GROUP_SIZE 1042 max_work_group_size = CL_DEVICE_MAX_WORK_GROUP_SIZE
363 image2d_max_width = CL_DEVICE_IMAGE2D_MAX_WIDTH 1043 image2d_max_width = CL_DEVICE_IMAGE2D_MAX_WIDTH
364 image2d_max_height = CL_DEVICE_IMAGE2D_MAX_HEIGHT 1044 image2d_max_height = CL_DEVICE_IMAGE2D_MAX_HEIGHT
365 image3d_max_width = CL_DEVICE_IMAGE3D_MAX_WIDTH 1045 image3d_max_width = CL_DEVICE_IMAGE3D_MAX_WIDTH
367 image3d_max_depth = CL_DEVICE_IMAGE3D_MAX_DEPTH 1047 image3d_max_depth = CL_DEVICE_IMAGE3D_MAX_DEPTH
368 max_parameter_size = CL_DEVICE_MAX_PARAMETER_SIZE 1048 max_parameter_size = CL_DEVICE_MAX_PARAMETER_SIZE
369 profiling_timer_resolution = CL_DEVICE_PROFILING_TIMER_RESOLUTION 1049 profiling_timer_resolution = CL_DEVICE_PROFILING_TIMER_RESOLUTION
370 PPCODE: 1050 PPCODE:
371 size_t value [1]; 1051 size_t value [1];
372 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1052 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
373 EXTEND (SP, 1); 1053 EXTEND (SP, 1);
374 const int i = 0; 1054 const int i = 0;
375 PUSHs (sv_2mortal (newSVuv (value [i]))); 1055 PUSHs (sv_2mortal (newSVuv (value [i])));
376 1056
377void 1057void
378max_work_item_sizes (OpenCL::Device this) 1058max_work_item_sizes (OpenCL::Device self)
379 PPCODE: 1059 PPCODE:
380 size_t size; 1060 size_t size;
381 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_MAX_WORK_ITEM_SIZES, 0, 0, &size)); 1061 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_MAX_WORK_ITEM_SIZES, 0, 0, &size));
382 size_t *value = tmpbuf (size); 1062 size_t *value = tmpbuf (size);
383 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_MAX_WORK_ITEM_SIZES, size, value, 0)); 1063 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_MAX_WORK_ITEM_SIZES, size, value, 0));
384 int i, n = size / sizeof (*value); 1064 int i, n = size / sizeof (*value);
385 EXTEND (SP, n); 1065 EXTEND (SP, n);
386 for (i = 0; i < n; ++i) 1066 for (i = 0; i < n; ++i)
387 PUSHs (sv_2mortal (newSVuv (value [i]))); 1067 PUSHs (sv_2mortal (newSVuv (value [i])));
388 1068
389void 1069void
390address_bits (OpenCL::Device this) 1070address_bits (OpenCL::Device self)
391 PPCODE: 1071 PPCODE:
392 cl_bitfield value [1]; 1072 cl_bitfield value [1];
393 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_ADDRESS_BITS, sizeof (value), value, 0)); 1073 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_ADDRESS_BITS, sizeof (value), value, 0));
394 EXTEND (SP, 1); 1074 EXTEND (SP, 1);
395 const int i = 0; 1075 const int i = 0;
396 PUSHs (sv_2mortal (newSVuv (value [i]))); 1076 PUSHs (sv_2mortal (newSVuv (value [i])));
397 1077
398void 1078void
399max_mem_alloc_size (OpenCL::Device this) 1079max_mem_alloc_size (OpenCL::Device self)
400 ALIAS: 1080 ALIAS:
401 max_mem_alloc_size = CL_DEVICE_MAX_MEM_ALLOC_SIZE 1081 max_mem_alloc_size = CL_DEVICE_MAX_MEM_ALLOC_SIZE
402 global_mem_cache_size = CL_DEVICE_GLOBAL_MEM_CACHE_SIZE 1082 global_mem_cache_size = CL_DEVICE_GLOBAL_MEM_CACHE_SIZE
403 global_mem_size = CL_DEVICE_GLOBAL_MEM_SIZE 1083 global_mem_size = CL_DEVICE_GLOBAL_MEM_SIZE
404 max_constant_buffer_size = CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE 1084 max_constant_buffer_size = CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE
405 local_mem_size = CL_DEVICE_LOCAL_MEM_SIZE 1085 local_mem_size = CL_DEVICE_LOCAL_MEM_SIZE
406 PPCODE: 1086 PPCODE:
407 cl_ulong value [1]; 1087 cl_ulong value [1];
408 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1088 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
409 EXTEND (SP, 1); 1089 EXTEND (SP, 1);
410 const int i = 0; 1090 const int i = 0;
411 PUSHs (sv_2mortal (newSVuv (value [i]))); 1091 PUSHs (sv_2mortal (newSVuv (value [i])));
412 1092
413void 1093void
414single_fp_config (OpenCL::Device this) 1094single_fp_config (OpenCL::Device self)
415 ALIAS: 1095 ALIAS:
416 single_fp_config = CL_DEVICE_SINGLE_FP_CONFIG 1096 single_fp_config = CL_DEVICE_SINGLE_FP_CONFIG
417 double_fp_config = CL_DEVICE_DOUBLE_FP_CONFIG 1097 double_fp_config = CL_DEVICE_DOUBLE_FP_CONFIG
418 half_fp_config = CL_DEVICE_HALF_FP_CONFIG 1098 half_fp_config = CL_DEVICE_HALF_FP_CONFIG
419 PPCODE: 1099 PPCODE:
420 cl_device_fp_config value [1]; 1100 cl_device_fp_config value [1];
421 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1101 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
422 EXTEND (SP, 1); 1102 EXTEND (SP, 1);
423 const int i = 0; 1103 const int i = 0;
424 PUSHs (sv_2mortal (newSVuv (value [i]))); 1104 PUSHs (sv_2mortal (newSVuv (value [i])));
425 1105
426void 1106void
427global_mem_cache_type (OpenCL::Device this) 1107global_mem_cache_type (OpenCL::Device self)
428 PPCODE: 1108 PPCODE:
429 cl_device_mem_cache_type value [1]; 1109 cl_device_mem_cache_type value [1];
430 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, sizeof (value), value, 0)); 1110 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_GLOBAL_MEM_CACHE_TYPE, sizeof (value), value, 0));
431 EXTEND (SP, 1); 1111 EXTEND (SP, 1);
432 const int i = 0; 1112 const int i = 0;
433 PUSHs (sv_2mortal (newSVuv (value [i]))); 1113 PUSHs (sv_2mortal (newSVuv (value [i])));
434 1114
435void 1115void
436local_mem_type (OpenCL::Device this) 1116local_mem_type (OpenCL::Device self)
437 PPCODE: 1117 PPCODE:
438 cl_device_local_mem_type value [1]; 1118 cl_device_local_mem_type value [1];
439 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_LOCAL_MEM_TYPE, sizeof (value), value, 0)); 1119 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_LOCAL_MEM_TYPE, sizeof (value), value, 0));
440 EXTEND (SP, 1); 1120 EXTEND (SP, 1);
441 const int i = 0; 1121 const int i = 0;
442 PUSHs (sv_2mortal (newSVuv (value [i]))); 1122 PUSHs (sv_2mortal (newSVuv (value [i])));
443 1123
444void 1124void
445error_correction_support (OpenCL::Device this) 1125error_correction_support (OpenCL::Device self)
446 ALIAS: 1126 ALIAS:
447 error_correction_support = CL_DEVICE_ERROR_CORRECTION_SUPPORT 1127 error_correction_support = CL_DEVICE_ERROR_CORRECTION_SUPPORT
448 endian_little = CL_DEVICE_ENDIAN_LITTLE 1128 endian_little = CL_DEVICE_ENDIAN_LITTLE
449 available = CL_DEVICE_AVAILABLE 1129 available = CL_DEVICE_AVAILABLE
450 compiler_available = CL_DEVICE_COMPILER_AVAILABLE 1130 compiler_available = CL_DEVICE_COMPILER_AVAILABLE
451 host_unified_memory = CL_DEVICE_HOST_UNIFIED_MEMORY 1131 host_unified_memory = CL_DEVICE_HOST_UNIFIED_MEMORY
452 PPCODE: 1132 PPCODE:
453 cl_bool value [1]; 1133 cl_bool value [1];
454 NEED_SUCCESS (GetDeviceInfo, (this, ix, sizeof (value), value, 0)); 1134 NEED_SUCCESS (GetDeviceInfo, (self, ix, sizeof (value), value, 0));
455 EXTEND (SP, 1); 1135 EXTEND (SP, 1);
456 const int i = 0; 1136 const int i = 0;
457 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no)); 1137 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no));
458 1138
459void 1139void
460execution_capabilities (OpenCL::Device this) 1140execution_capabilities (OpenCL::Device self)
461 PPCODE: 1141 PPCODE:
462 cl_device_exec_capabilities value [1]; 1142 cl_device_exec_capabilities value [1];
463 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_EXECUTION_CAPABILITIES, sizeof (value), value, 0)); 1143 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_EXECUTION_CAPABILITIES, sizeof (value), value, 0));
464 EXTEND (SP, 1); 1144 EXTEND (SP, 1);
465 const int i = 0; 1145 const int i = 0;
466 PUSHs (sv_2mortal (newSVuv (value [i]))); 1146 PUSHs (sv_2mortal (newSVuv (value [i])));
467 1147
468void 1148void
469properties (OpenCL::Device this) 1149properties (OpenCL::Device self)
470 PPCODE: 1150 PPCODE:
471 cl_command_queue_properties value [1]; 1151 cl_command_queue_properties value [1];
472 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0)); 1152 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_QUEUE_PROPERTIES, sizeof (value), value, 0));
473 EXTEND (SP, 1); 1153 EXTEND (SP, 1);
474 const int i = 0; 1154 const int i = 0;
475 PUSHs (sv_2mortal (newSViv (value [i]))); 1155 PUSHs (sv_2mortal (newSVuv (value [i])));
476 1156
477void 1157void
478platform (OpenCL::Device this) 1158platform (OpenCL::Device self)
479 PPCODE: 1159 PPCODE:
480 cl_platform_id value [1]; 1160 cl_platform_id value [1];
481 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_PLATFORM, sizeof (value), value, 0)); 1161 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PLATFORM, sizeof (value), value, 0));
482 EXTEND (SP, 1); 1162 EXTEND (SP, 1);
483 const int i = 0; 1163 const int i = 0;
484 { 1164 {
485 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Platform", value [i])); 1165 PUSH_CLOBJ (stash_platform, value [i]);
486 } 1166 }
487 1167
488void 1168void
489name (OpenCL::Device this) 1169name (OpenCL::Device self)
490 ALIAS: 1170 ALIAS:
491 name = CL_DEVICE_NAME 1171 name = CL_DEVICE_NAME
492 vendor = CL_DEVICE_VENDOR 1172 vendor = CL_DEVICE_VENDOR
493 driver_version = CL_DRIVER_VERSION 1173 driver_version = CL_DRIVER_VERSION
494 profile = CL_DEVICE_PROFILE 1174 profile = CL_DEVICE_PROFILE
495 version = CL_DEVICE_VERSION 1175 version = CL_DEVICE_VERSION
496 extensions = CL_DEVICE_EXTENSIONS 1176 extensions = CL_DEVICE_EXTENSIONS
497 PPCODE: 1177 PPCODE:
498 size_t size; 1178 size_t size;
499 NEED_SUCCESS (GetDeviceInfo, (this, ix, 0, 0, &size)); 1179 NEED_SUCCESS (GetDeviceInfo, (self, ix, 0, 0, &size));
500 char *value = tmpbuf (size); 1180 char *value = tmpbuf (size);
501 NEED_SUCCESS (GetDeviceInfo, (this, ix, size, value, 0)); 1181 NEED_SUCCESS (GetDeviceInfo, (self, ix, size, value, 0));
502 EXTEND (SP, 1); 1182 EXTEND (SP, 1);
503 const int i = 0; 1183 const int i = 0;
504 PUSHs (sv_2mortal (newSVpv (value, 0))); 1184 PUSHs (sv_2mortal (newSVpv (value, 0)));
505 1185
506void 1186void
507parent_device_ext (OpenCL::Device this) 1187parent_device_ext (OpenCL::Device self)
508 PPCODE: 1188 PPCODE:
509 cl_device_id value [1]; 1189 cl_device_id value [1];
510 NEED_SUCCESS (GetDeviceInfo, (this, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0)); 1190 NEED_SUCCESS (GetDeviceInfo, (self, CL_DEVICE_PARENT_DEVICE_EXT, sizeof (value), value, 0));
511 EXTEND (SP, 1); 1191 EXTEND (SP, 1);
512 const int i = 0; 1192 const int i = 0;
513 { 1193 {
514 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 1194 PUSH_CLOBJ (stash_device, value [i]);
515 } 1195 }
516 1196
517void 1197void
518partition_types_ext (OpenCL::Device this) 1198partition_types_ext (OpenCL::Device self)
519 ALIAS: 1199 ALIAS:
520 partition_types_ext = CL_DEVICE_PARTITION_TYPES_EXT 1200 partition_types_ext = CL_DEVICE_PARTITION_TYPES_EXT
521 affinity_domains_ext = CL_DEVICE_AFFINITY_DOMAINS_EXT 1201 affinity_domains_ext = CL_DEVICE_AFFINITY_DOMAINS_EXT
522 partition_style_ext = CL_DEVICE_PARTITION_STYLE_EXT 1202 partition_style_ext = CL_DEVICE_PARTITION_STYLE_EXT
523 PPCODE: 1203 PPCODE:
524 size_t size; 1204 size_t size;
525 NEED_SUCCESS (GetDeviceInfo, (this, ix, 0, 0, &size)); 1205 NEED_SUCCESS (GetDeviceInfo, (self, ix, 0, 0, &size));
526 cl_device_partition_property_ext *value = tmpbuf (size); 1206 cl_device_partition_property_ext *value = tmpbuf (size);
527 NEED_SUCCESS (GetDeviceInfo, (this, ix, size, value, 0)); 1207 NEED_SUCCESS (GetDeviceInfo, (self, ix, size, value, 0));
528 int i, n = size / sizeof (*value); 1208 int i, n = size / sizeof (*value);
529 EXTEND (SP, n); 1209 EXTEND (SP, n);
530 for (i = 0; i < n; ++i) 1210 for (i = 0; i < n; ++i)
531 PUSHs (sv_2mortal (newSVuv (value [i]))); 1211 PUSHs (sv_2mortal (newSVuv (value [i])));
532 1212
533#END:device 1213#END:device
534 1214
1215MODULE = OpenCL PACKAGE = OpenCL::SubDevice
1216
1217#if CL_VERSION_1_2
1218
1219void
1220DESTROY (OpenCL::SubDevice self)
1221 CODE:
1222 clReleaseDevice (self);
1223
1224#endif
1225
535MODULE = OpenCL PACKAGE = OpenCL::Context 1226MODULE = OpenCL PACKAGE = OpenCL::Context
536 1227
537void 1228void
538DESTROY (OpenCL::Context context) 1229DESTROY (OpenCL::Context self)
539 CODE: 1230 CODE:
540 clReleaseContext (context); 1231 clReleaseContext (self);
541 1232
542void 1233void
543info (OpenCL::Context this, cl_context_info name) 1234info (OpenCL::Context self, cl_context_info name)
544 PPCODE: 1235 PPCODE:
545 INFO (Context) 1236 INFO (Context)
546 1237
547void 1238void
548queue (OpenCL::Context this, OpenCL::Device device, cl_command_queue_properties properties = 0) 1239queue (OpenCL::Context self, OpenCL::Device device, cl_command_queue_properties properties = 0)
549 PPCODE: 1240 PPCODE:
550 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (this, device, properties, &res)); 1241 NEED_SUCCESS_ARG (cl_command_queue queue, CreateCommandQueue, (self, device, properties, &res));
551 XPUSH_NEW_OBJ ("OpenCL::Queue", queue); 1242 XPUSH_CLOBJ (stash_queue, queue);
552 1243
553void 1244void
554user_event (OpenCL::Context this) 1245user_event (OpenCL::Context self)
555 PPCODE: 1246 PPCODE:
556 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (this, &res)); 1247 NEED_SUCCESS_ARG (cl_event ev, CreateUserEvent, (self, &res));
557 XPUSH_NEW_OBJ ("OpenCL::UserEvent", ev); 1248 XPUSH_CLOBJ (stash_userevent, ev);
558 1249
559void 1250void
560buffer (OpenCL::Context this, cl_mem_flags flags, size_t len) 1251buffer (OpenCL::Context self, cl_mem_flags flags, size_t len)
561 PPCODE: 1252 PPCODE:
562 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)) 1253 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))
563 croak ("clCreateBuffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?"); 1254 croak ("OpenCL::Context::buffer: cannot use/copy host ptr when no data is given, use $context->buffer_sv instead?");
564 1255
565 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (this, flags, len, 0, &res)); 1256 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, 0, &res));
566 XPUSH_NEW_OBJ ("OpenCL::BufferObj", mem); 1257 XPUSH_CLOBJ (stash_bufferobj, mem);
567 1258
568void 1259void
569buffer_sv (OpenCL::Context this, cl_mem_flags flags, SV *data) 1260buffer_sv (OpenCL::Context self, cl_mem_flags flags, SV *data)
570 PPCODE: 1261 PPCODE:
571 STRLEN len; 1262 STRLEN len;
572 char *ptr = SvPVbyte (data, len); 1263 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
573
574 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR))) 1264 if (!(flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR)))
575 croak ("clCreateBuffer: have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?"); 1265 croak ("OpenCL::Context::buffer_sv: you have to specify use or copy host ptr when buffer data is given, use $context->buffer instead?");
576
577 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (this, flags, len, ptr, &res)); 1266 NEED_SUCCESS_ARG (cl_mem mem, CreateBuffer, (self, flags, len, ptr, &res));
578 XPUSH_NEW_OBJ ("OpenCL::BufferObj", mem); 1267 XPUSH_CLOBJ (stash_bufferobj, mem);
579 1268
1269#if CL_VERSION_1_2
1270
580void 1271void
581image2d (OpenCL::Context this, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, size_t width, size_t height, size_t row_pitch = 0, SV *data = &PL_sv_undef) 1272image (OpenCL::Context self, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, cl_mem_object_type type, size_t width, size_t height, size_t depth = 0, size_t array_size = 0, size_t row_pitch = 0, size_t slice_pitch = 0, cl_uint num_mip_level = 0, cl_uint num_samples = 0, SV *data = &PL_sv_undef)
582 PPCODE: 1273 PPCODE:
583 STRLEN len; 1274 STRLEN len;
584 char *ptr = SvPVbyte (data, len); 1275 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
585 const cl_image_format format = { channel_order, channel_type }; 1276 const cl_image_format format = { channel_order, channel_type };
1277 const cl_image_desc desc = {
1278 type,
1279 width, height, depth,
1280 array_size, row_pitch, slice_pitch,
1281 num_mip_level, num_samples,
1282 type == CL_MEM_OBJECT_IMAGE1D_BUFFER ? (cl_mem)SvCLOBJ (cv, "data", data, "OpenCL::Buffer") : 0
1283 };
586 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (this, flags, &format, width, height, row_pitch, ptr, &res)); 1284 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
587 XPUSH_NEW_OBJ ("OpenCL::Image2D", mem); 1285 HV *stash = stash_image;
1286 switch (type)
1287 {
1288 case CL_MEM_OBJECT_IMAGE1D_BUFFER: stash = stash_image1dbuffer; break;
1289 case CL_MEM_OBJECT_IMAGE1D: stash = stash_image1d; break;
1290 case CL_MEM_OBJECT_IMAGE1D_ARRAY: stash = stash_image2darray; break;
1291 case CL_MEM_OBJECT_IMAGE2D: stash = stash_image2d; break;
1292 case CL_MEM_OBJECT_IMAGE2D_ARRAY: stash = stash_image2darray; break;
1293 case CL_MEM_OBJECT_IMAGE3D: stash = stash_image3d; break;
1294 }
1295 XPUSH_CLOBJ (stash, mem);
588 1296
1297#endif
1298
589void 1299void
590image3d (OpenCL::Context this, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, size_t width, size_t height, size_t depth, size_t row_pitch = 0, size_t slice_pitch = 0, SV *data = &PL_sv_undef) 1300image2d (OpenCL::Context self, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, size_t width, size_t height, size_t row_pitch = 0, SV *data = &PL_sv_undef)
591 PPCODE: 1301 PPCODE:
592 STRLEN len; 1302 STRLEN len;
593 char *ptr = SvPVbyte (data, len); 1303 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
594 const cl_image_format format = { channel_order, channel_type }; 1304 const cl_image_format format = { channel_order, channel_type };
1305#if PREFER_1_1
1306 NEED_SUCCESS_ARG (cl_mem mem, CreateImage2D, (self, flags, &format, width, height, row_pitch, ptr, &res));
1307#else
1308 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE2D, width, height, 0, 0, row_pitch, 0, 0, 0, 0 };
1309 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1310#endif
1311 XPUSH_CLOBJ (stash_image2d, mem);
1312
1313void
1314image3d (OpenCL::Context self, cl_mem_flags flags, cl_channel_order channel_order, cl_channel_type channel_type, size_t width, size_t height, size_t depth, size_t row_pitch = 0, size_t slice_pitch = 0, SV *data = &PL_sv_undef)
1315 PPCODE:
1316 STRLEN len;
1317 char *ptr = SvOK (data) ? SvPVbyte (data, len) : 0;
1318 const cl_image_format format = { channel_order, channel_type };
1319#if PREFER_1_1
595 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (this, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res)); 1320 NEED_SUCCESS_ARG (cl_mem mem, CreateImage3D, (self, flags, &format, width, height, depth, row_pitch, slice_pitch, ptr, &res));
596 XPUSH_NEW_OBJ ("OpenCL::Image3D", mem); 1321#else
1322 const cl_image_desc desc = { CL_MEM_OBJECT_IMAGE3D, width, height, depth, 0, row_pitch, slice_pitch, 0, 0, 0 };
1323 NEED_SUCCESS_ARG (cl_mem mem, CreateImage, (self, flags, &format, &desc, ptr, &res));
1324#endif
1325 XPUSH_CLOBJ (stash_image3d, mem);
597 1326
1327#if cl_apple_gl_sharing || cl_khr_gl_sharing
1328
598void 1329void
1330gl_buffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint bufobj)
1331 PPCODE:
1332 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLBuffer, (self, flags, bufobj, &res));
1333 XPUSH_CLOBJ (stash_bufferobj, mem);
1334
1335void
1336gl_renderbuffer (OpenCL::Context self, cl_mem_flags flags, cl_GLuint renderbuffer)
1337 PPCODE:
1338 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLRenderbuffer, (self, flags, renderbuffer, &res));
1339 XPUSH_CLOBJ (stash_image2d, mem);
1340
1341#if CL_VERSION_1_2
1342
1343void
1344gl_texture (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1345 ALIAS:
1346 PPCODE:
1347 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture, (self, flags, target, miplevel, texture, &res));
1348 cl_gl_object_type type;
1349 NEED_SUCCESS (GetGLObjectInfo, (mem, &type, 0)); // TODO: use target instead?
1350 HV *stash = stash_memory;
1351 switch (type)
1352 {
1353 case CL_GL_OBJECT_TEXTURE_BUFFER: stash = stash_image1dbuffer; break;
1354 case CL_GL_OBJECT_TEXTURE1D: stash = stash_image1d; break;
1355 case CL_GL_OBJECT_TEXTURE1D_ARRAY: stash = stash_image2darray; break;
1356 case CL_GL_OBJECT_TEXTURE2D: stash = stash_image2d; break;
1357 case CL_GL_OBJECT_TEXTURE2D_ARRAY: stash = stash_image2darray; break;
1358 case CL_GL_OBJECT_TEXTURE3D: stash = stash_image3d; break;
1359 }
1360 XPUSH_CLOBJ (stash, mem);
1361
1362#endif
1363
1364void
1365gl_texture2d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1366 PPCODE:
1367#if PREFER_1_1
1368 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture2D, (self, flags, target, miplevel, texture, &res));
1369#else
1370 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1371#endif
1372 XPUSH_CLOBJ (stash_image2d, mem);
1373
1374void
1375gl_texture3d (OpenCL::Context self, cl_mem_flags flags, cl_GLenum target, cl_GLint miplevel, cl_GLuint texture)
1376 PPCODE:
1377#if PREFER_1_1
1378 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture3D, (self, flags, target, miplevel, texture, &res));
1379#else
1380 NEED_SUCCESS_ARG (cl_mem mem, CreateFromGLTexture , (self, flags, target, miplevel, texture, &res));
1381#endif
1382 XPUSH_CLOBJ (stash_image3d, mem);
1383
1384#endif
1385
1386void
599supported_image_formats (OpenCL::Context this, cl_mem_flags flags, cl_mem_object_type image_type) 1387supported_image_formats (OpenCL::Context self, cl_mem_flags flags, cl_mem_object_type image_type)
600 PPCODE: 1388 PPCODE:
601{ 1389{
602 cl_uint count; 1390 cl_uint count;
603 cl_image_format *list; 1391 cl_image_format *list;
604 int i; 1392 int i;
605 1393
606 NEED_SUCCESS (GetSupportedImageFormats, (this, flags, image_type, 0, 0, &count)); 1394 NEED_SUCCESS (GetSupportedImageFormats, (self, flags, image_type, 0, 0, &count));
607 Newx (list, count, cl_image_format); 1395 Newx (list, count, cl_image_format);
608 NEED_SUCCESS (GetSupportedImageFormats, (this, flags, image_type, count, list, 0)); 1396 NEED_SUCCESS (GetSupportedImageFormats, (self, flags, image_type, count, list, 0));
609 1397
610 EXTEND (SP, count); 1398 EXTEND (SP, count);
611 for (i = 0; i < count; ++i) 1399 for (i = 0; i < count; ++i)
612 { 1400 {
613 AV *av = newAV (); 1401 AV *av = newAV ();
616 PUSHs (sv_2mortal (newRV_noinc ((SV *)av))); 1404 PUSHs (sv_2mortal (newRV_noinc ((SV *)av)));
617 } 1405 }
618} 1406}
619 1407
620void 1408void
621sampler (OpenCL::Context this, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode) 1409sampler (OpenCL::Context self, cl_bool normalized_coords, cl_addressing_mode addressing_mode, cl_filter_mode filter_mode)
622 PPCODE: 1410 PPCODE:
623 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (this, normalized_coords, addressing_mode, filter_mode, &res)); 1411 NEED_SUCCESS_ARG (cl_sampler sampler, CreateSampler, (self, normalized_coords, addressing_mode, filter_mode, &res));
624 XPUSH_NEW_OBJ ("OpenCL::Sampler", sampler); 1412 XPUSH_CLOBJ (stash_sampler, sampler);
625 1413
626void 1414void
627program_with_source (OpenCL::Context this, SV *program) 1415program_with_source (OpenCL::Context self, SV *program)
628 PPCODE: 1416 PPCODE:
629 STRLEN len; 1417 STRLEN len;
630 size_t len2; 1418 size_t len2;
631 const char *ptr = SvPVbyte (program, len); 1419 const char *ptr = SvPVbyte (program, len);
632 1420
633 len2 = len; 1421 len2 = len;
634 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (this, 1, &ptr, &len2, &res)); 1422 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithSource, (self, 1, &ptr, &len2, &res));
635 XPUSH_NEW_OBJ ("OpenCL::Program", prog); 1423 XPUSH_CLOBJ (stash_program, prog);
1424
1425void
1426program_with_binary (OpenCL::Context self, SV *devices, SV *binaries)
1427 PPCODE:
1428 cl_uint device_count;
1429 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
1430
1431 if (!SvROK (binaries) || SvTYPE (SvRV (binaries)) != SVt_PVAV)
1432 croak ("OpenCL::Context::program_with_binary: binaries must be specified as reference to an array of strings");
1433
1434 binaries = SvRV (binaries);
1435
1436 if (device_count != av_len ((AV *)binaries) + 1)
1437 croak ("OpenCL::Context::program_with_binary: differing numbers of devices and binaries are not allowed");
1438
1439 size_t *length_list = tmpbuf (sizeof (*length_list) * device_count);
1440 const unsigned char **binary_list = tmpbuf (sizeof (*binary_list) * device_count);
1441 cl_int *status_list = tmpbuf (sizeof (*status_list) * device_count);
1442
1443 int i;
1444 for (i = 0; i < device_count; ++i)
1445 {
1446 STRLEN len;
1447 binary_list [i] = (const unsigned char *)SvPVbyte (*av_fetch ((AV *)binaries, i, 0), len);
1448 length_list [i] = len;
1449 }
1450
1451 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithBinary, (self, device_count, device_list,
1452 length_list, binary_list,
1453 GIMME_V == G_ARRAY ? status_list : 0, &res));
1454
1455 EXTEND (SP, 2);
1456 PUSH_CLOBJ (stash_program, prog);
1457
1458 if (GIMME_V == G_ARRAY)
1459 {
1460 AV *av = newAV ();
1461 PUSHs (sv_2mortal (newRV_noinc ((SV *)av)));
1462
1463 for (i = device_count; i--; )
1464 av_store (av, i, newSViv (status_list [i]));
1465 }
1466
1467#if CL_VERSION_1_2
1468
1469void
1470program_with_built_in_kernels (OpenCL::Context self, SV *devices, SV *kernel_names)
1471 PPCODE:
1472 cl_uint device_count;
1473 cl_device_id *device_list = object_list (cv, 0, "devices", devices, "OpenCL::Device", &device_count);
1474
1475 NEED_SUCCESS_ARG (cl_program prog, CreateProgramWithBuiltInKernels, (self, device_count, device_list, SvPVbyte_nolen (kernel_names), &res));
1476
1477 XPUSH_CLOBJ (stash_program, prog);
1478
1479void
1480link_program (OpenCL::Context self, SV *devices, SV *options, SV *programs, SV *notify = &PL_sv_undef)
1481 CODE:
1482 cl_uint device_count = 0;
1483 cl_device_id *device_list = 0;
1484
1485 if (SvOK (devices))
1486 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
1487
1488 cl_uint program_count;
1489 cl_program *program_list = object_list (cv, 0, "programs", programs, "OpenCL::Program", &program_count);
1490
1491 void *user_data;
1492 program_callback pfn_notify = make_program_callback (notify, &user_data);
1493
1494 NEED_SUCCESS_ARG (cl_program prog, LinkProgram, (self, device_count, device_list, SvPVbyte_nolen (options),
1495 program_count, program_list, pfn_notify, user_data, &res));
1496
1497 XPUSH_CLOBJ (stash_program, prog);
1498
1499#endif
636 1500
637#BEGIN:context 1501#BEGIN:context
638 1502
639void 1503void
640reference_count (OpenCL::Context this) 1504reference_count (OpenCL::Context self)
641 ALIAS: 1505 ALIAS:
642 reference_count = CL_CONTEXT_REFERENCE_COUNT 1506 reference_count = CL_CONTEXT_REFERENCE_COUNT
643 num_devices = CL_CONTEXT_NUM_DEVICES 1507 num_devices = CL_CONTEXT_NUM_DEVICES
644 PPCODE: 1508 PPCODE:
645 cl_uint value [1]; 1509 cl_uint value [1];
646 NEED_SUCCESS (GetContextInfo, (this, ix, sizeof (value), value, 0)); 1510 NEED_SUCCESS (GetContextInfo, (self, ix, sizeof (value), value, 0));
647 EXTEND (SP, 1); 1511 EXTEND (SP, 1);
648 const int i = 0; 1512 const int i = 0;
649 PUSHs (sv_2mortal (newSVuv (value [i]))); 1513 PUSHs (sv_2mortal (newSVuv (value [i])));
650 1514
651void 1515void
652devices (OpenCL::Context this) 1516devices (OpenCL::Context self)
653 PPCODE: 1517 PPCODE:
654 size_t size; 1518 size_t size;
655 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_DEVICES, 0, 0, &size)); 1519 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, 0, 0, &size));
656 cl_device_id *value = tmpbuf (size); 1520 cl_device_id *value = tmpbuf (size);
657 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_DEVICES, size, value, 0)); 1521 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_DEVICES, size, value, 0));
658 int i, n = size / sizeof (*value); 1522 int i, n = size / sizeof (*value);
659 EXTEND (SP, n); 1523 EXTEND (SP, n);
660 for (i = 0; i < n; ++i) 1524 for (i = 0; i < n; ++i)
661 { 1525 {
662 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 1526 PUSH_CLOBJ (stash_device, value [i]);
663 } 1527 }
664 1528
665void 1529void
666properties (OpenCL::Context this) 1530properties (OpenCL::Context self)
667 PPCODE: 1531 PPCODE:
668 size_t size; 1532 size_t size;
669 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_PROPERTIES, 0, 0, &size)); 1533 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_PROPERTIES, 0, 0, &size));
670 cl_context_properties *value = tmpbuf (size); 1534 cl_context_properties *value = tmpbuf (size);
671 NEED_SUCCESS (GetContextInfo, (this, CL_CONTEXT_PROPERTIES, size, value, 0)); 1535 NEED_SUCCESS (GetContextInfo, (self, CL_CONTEXT_PROPERTIES, size, value, 0));
672 int i, n = size / sizeof (*value); 1536 int i, n = size / sizeof (*value);
673 EXTEND (SP, n); 1537 EXTEND (SP, n);
674 for (i = 0; i < n; ++i) 1538 for (i = 0; i < n; ++i)
675 PUSHs (sv_2mortal (newSVuv ((UV)value [i]))); 1539 PUSHs (sv_2mortal (newSVuv ((UV)value [i])));
676 1540
677#END:context 1541#END:context
678 1542
679MODULE = OpenCL PACKAGE = OpenCL::Queue 1543MODULE = OpenCL PACKAGE = OpenCL::Queue
680 1544
681void 1545void
682DESTROY (OpenCL::Queue this) 1546DESTROY (OpenCL::Queue self)
683 CODE: 1547 CODE:
684 clReleaseCommandQueue (this); 1548 clReleaseCommandQueue (self);
685 1549
686void 1550void
687enqueue_read_buffer (OpenCL::Queue this, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...) 1551read_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, size_t len, SV *data, ...)
1552 ALIAS:
1553 enqueue_read_buffer = 0
688 PPCODE: 1554 PPCODE:
689 cl_event ev = 0;
690 EVENT_LIST (6, items - 6); 1555 EVENT_LIST (6);
691 1556
692 SvUPGRADE (data, SVt_PV); 1557 SvUPGRADE (data, SVt_PV);
693 SvGROW (data, len); 1558 SvGROW (data, len);
694 SvPOK_only (data); 1559 SvPOK_only (data);
695 SvCUR_set (data, len); 1560 SvCUR_set (data, len);
1561
1562 cl_event ev = 0;
696 NEED_SUCCESS (EnqueueReadBuffer, (this, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1563 NEED_SUCCESS (EnqueueReadBuffer, (self, mem, blocking, offset, len, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
697 1564
698 if (ev) 1565 if (ev)
699 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1566 XPUSH_CLOBJ (stash_event, ev);
700 1567
701void 1568void
702enqueue_write_buffer (OpenCL::Queue this, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...) 1569write_buffer (OpenCL::Queue self, OpenCL::Buffer mem, cl_bool blocking, size_t offset, SV *data, ...)
1570 ALIAS:
1571 enqueue_write_buffer = 0
703 PPCODE: 1572 PPCODE:
704 cl_event ev = 0; 1573 EVENT_LIST (5);
1574
705 STRLEN len; 1575 STRLEN len;
706 char *ptr = SvPVbyte (data, len); 1576 char *ptr = SvPVbyte (data, len);
707 EVENT_LIST (5, items - 5);
708 1577
1578 cl_event ev = 0;
709 NEED_SUCCESS (EnqueueReadBuffer, (this, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1579 NEED_SUCCESS (EnqueueWriteBuffer, (self, mem, blocking, offset, len, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
710 1580
711 if (ev) 1581 if (ev)
712 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1582 XPUSH_CLOBJ (stash_event, ev);
713 1583
1584#if CL_VERSION_1_2
1585
714void 1586void
715enqueue_copy_buffer (OpenCL::Queue this, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...) 1587fill_buffer (OpenCL::Queue self, OpenCL::Buffer mem, SV *data, size_t offset, size_t size, ...)
1588 ALIAS:
1589 enqueue_fill_buffer = 0
716 PPCODE: 1590 PPCODE:
1591 EVENT_LIST (5);
1592
1593 STRLEN len;
1594 char *ptr = SvPVbyte (data, len);
1595
717 cl_event ev = 0; 1596 cl_event ev = 0;
718 EVENT_LIST (6, items - 6);
719
720 NEED_SUCCESS (EnqueueCopyBuffer, (this, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1597 NEED_SUCCESS (EnqueueFillBuffer, (self, mem, ptr, len, offset, size, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
721 1598
722 if (ev) 1599 if (ev)
723 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1600 XPUSH_CLOBJ (stash_event, ev);
724 1601
725void 1602void
726enqueue_read_buffer_rect (OpenCL::Queue this, OpenCL::Memory buf, cl_bool blocking, size_t buf_x, size_t buf_y, size_t buf_z, size_t host_x, size_t host_y, size_t host_z, size_t width, size_t height, size_t depth, size_t buf_row_pitch, size_t buf_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, SV *data, ...) 1603fill_image (OpenCL::Queue self, OpenCL::Image img, NV r, NV g, NV b, NV a, size_t x, size_t y, size_t z, size_t width, size_t height, size_t depth, ...)
1604 ALIAS:
1605 enqueue_fill_image = 0
727 PPCODE: 1606 PPCODE:
1607 EVENT_LIST (12);
1608
1609 const size_t origin [3] = { x, y, z };
1610 const size_t region [3] = { width, height, depth };
1611
1612 const cl_float c_f [4] = { r, g, b, a };
1613 const cl_uint c_u [4] = { r, g, b, a };
1614 const cl_int c_s [4] = { r, g, b, a };
1615 const void *c_fus [3] = { &c_f, &c_u, &c_s };
1616 static const char fus [] = { 0, 0, 0, 0, 0, 0, 0, 2, 2, 2, 1, 1, 1, 0, 0 };
1617 cl_image_format format;
1618 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
1619 assert (sizeof (fus) == CL_FLOAT + 1 - CL_SNORM_INT8);
1620 if (format.image_channel_data_type < CL_SNORM_INT8 || CL_FLOAT < format.image_channel_data_type)
1621 croak ("enqueue_fill_image: image has unsupported channel type, only opencl 1.2 channel types supported.");
1622
728 cl_event ev = 0; 1623 cl_event ev = 0;
1624 NEED_SUCCESS (EnqueueFillImage, (self, img, c_fus [fus [format.image_channel_data_type - CL_SNORM_INT8]],
1625 origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1626
1627 if (ev)
1628 XPUSH_CLOBJ (stash_event, ev);
1629
1630#endif
1631
1632void
1633copy_buffer (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_offset, size_t dst_offset, size_t len, ...)
1634 ALIAS:
1635 enqueue_copy_buffer = 0
1636 PPCODE:
1637 EVENT_LIST (6);
1638
1639 cl_event ev = 0;
1640 NEED_SUCCESS (EnqueueCopyBuffer, (self, src, dst, src_offset, dst_offset, len, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1641
1642 if (ev)
1643 XPUSH_CLOBJ (stash_event, ev);
1644
1645void
1646read_buffer_rect (OpenCL::Queue self, OpenCL::Memory buf, cl_bool blocking, size_t buf_x, size_t buf_y, size_t buf_z, size_t host_x, size_t host_y, size_t host_z, size_t width, size_t height, size_t depth, size_t buf_row_pitch, size_t buf_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, SV *data, ...)
1647 ALIAS:
1648 enqueue_read_buffer_rect = 0
1649 PPCODE:
1650 EVENT_LIST (17);
1651
729 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1652 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
730 const size_t host_origin[3] = { host_x, host_y, host_z }; 1653 const size_t host_origin[3] = { host_x, host_y, host_z };
731 const size_t region[3] = { width, height, depth }; 1654 const size_t region[3] = { width, height, depth };
732 EVENT_LIST (17, items - 17);
733 1655
734 if (!buf_row_pitch) 1656 if (!buf_row_pitch)
735 buf_row_pitch = region [0]; 1657 buf_row_pitch = region [0];
736 1658
737 if (!buf_slice_pitch) 1659 if (!buf_slice_pitch)
747 1669
748 SvUPGRADE (data, SVt_PV); 1670 SvUPGRADE (data, SVt_PV);
749 SvGROW (data, len); 1671 SvGROW (data, len);
750 SvPOK_only (data); 1672 SvPOK_only (data);
751 SvCUR_set (data, len); 1673 SvCUR_set (data, len);
1674
1675 cl_event ev = 0;
752 NEED_SUCCESS (EnqueueReadBufferRect, (this, buf, blocking, buf_origin, host_origin, region, buf_row_pitch, buf_slice_pitch, host_row_pitch, host_slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1676 NEED_SUCCESS (EnqueueReadBufferRect, (self, buf, blocking, buf_origin, host_origin, region, buf_row_pitch, buf_slice_pitch, host_row_pitch, host_slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
753 1677
754 if (ev) 1678 if (ev)
755 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1679 XPUSH_CLOBJ (stash_event, ev);
756 1680
757void 1681void
758enqueue_write_buffer_rect (OpenCL::Queue this, OpenCL::Memory buf, cl_bool blocking, size_t buf_x, size_t buf_y, size_t buf_z, size_t host_x, size_t host_y, size_t host_z, size_t width, size_t height, size_t depth, size_t buf_row_pitch, size_t buf_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, SV *data, ...) 1682write_buffer_rect (OpenCL::Queue self, OpenCL::Memory buf, cl_bool blocking, size_t buf_x, size_t buf_y, size_t buf_z, size_t host_x, size_t host_y, size_t host_z, size_t width, size_t height, size_t depth, size_t buf_row_pitch, size_t buf_slice_pitch, size_t host_row_pitch, size_t host_slice_pitch, SV *data, ...)
1683 ALIAS:
1684 enqueue_write_buffer_rect = 0
759 PPCODE: 1685 PPCODE:
760 cl_event ev = 0; 1686 EVENT_LIST (17);
1687
761 const size_t buf_origin [3] = { buf_x , buf_y , buf_z }; 1688 const size_t buf_origin [3] = { buf_x , buf_y , buf_z };
762 const size_t host_origin[3] = { host_x, host_y, host_z }; 1689 const size_t host_origin[3] = { host_x, host_y, host_z };
763 const size_t region[3] = { width, height, depth }; 1690 const size_t region[3] = { width, height, depth };
764 STRLEN len; 1691 STRLEN len;
765 char *ptr = SvPVbyte (data, len); 1692 char *ptr = SvPVbyte (data, len);
766 EVENT_LIST (17, items - 17);
767 1693
768 if (!buf_row_pitch) 1694 if (!buf_row_pitch)
769 buf_row_pitch = region [0]; 1695 buf_row_pitch = region [0];
770 1696
771 if (!buf_slice_pitch) 1697 if (!buf_slice_pitch)
780 size_t min_len = host_row_pitch * host_slice_pitch * region [2]; 1706 size_t min_len = host_row_pitch * host_slice_pitch * region [2];
781 1707
782 if (len < min_len) 1708 if (len < min_len)
783 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1709 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
784 1710
1711 cl_event ev = 0;
785 NEED_SUCCESS (EnqueueWriteBufferRect, (this, buf, blocking, buf_origin, host_origin, region, buf_row_pitch, buf_slice_pitch, host_row_pitch, host_slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1712 NEED_SUCCESS (EnqueueWriteBufferRect, (self, buf, blocking, buf_origin, host_origin, region, buf_row_pitch, buf_slice_pitch, host_row_pitch, host_slice_pitch, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
786 1713
787 if (ev) 1714 if (ev)
788 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1715 XPUSH_CLOBJ (stash_event, ev);
789 1716
790void 1717void
791enqueue_copy_buffer_rect (OpenCL::Queue this, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_x, size_t src_y, size_t src_z, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, size_t src_row_pitch, size_t src_slice_pitch, size_t dst_row_pitch, size_t dst_slice_pitch, ...) 1718copy_buffer_rect (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Buffer dst, size_t src_x, size_t src_y, size_t src_z, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, size_t src_row_pitch, size_t src_slice_pitch, size_t dst_row_pitch, size_t dst_slice_pitch, ...)
1719 ALIAS:
1720 enqueue_copy_buffer_rect = 0
792 PPCODE: 1721 PPCODE:
793 cl_event ev = 0; 1722 EVENT_LIST (16);
1723
794 const size_t src_origin[3] = { src_x, src_y, src_z }; 1724 const size_t src_origin[3] = { src_x, src_y, src_z };
795 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1725 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
796 const size_t region[3] = { width, height, depth }; 1726 const size_t region[3] = { width, height, depth };
797 EVENT_LIST (16, items - 16);
798 1727
1728 cl_event ev = 0;
799 NEED_SUCCESS (EnqueueCopyBufferRect, (this, src, dst, src_origin, dst_origin, region, src_row_pitch, src_slice_pitch, dst_row_pitch, dst_slice_pitch, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1729 NEED_SUCCESS (EnqueueCopyBufferRect, (self, src, dst, src_origin, dst_origin, region, src_row_pitch, src_slice_pitch, dst_row_pitch, dst_slice_pitch, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
800 1730
801 if (ev) 1731 if (ev)
802 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1732 XPUSH_CLOBJ (stash_event, ev);
803 1733
804void 1734void
805enqueue_read_image (OpenCL::Queue this, OpenCL::Image src, cl_bool blocking, size_t src_x, size_t src_y, size_t src_z, size_t width, size_t height, size_t depth, size_t row_pitch, size_t slice_pitch, SV *data, ...) 1735read_image (OpenCL::Queue self, OpenCL::Image src, cl_bool blocking, size_t src_x, size_t src_y, size_t src_z, size_t width, size_t height, size_t depth, size_t row_pitch, size_t slice_pitch, SV *data, ...)
1736 ALIAS:
1737 enqueue_read_image = 0
806 PPCODE: 1738 PPCODE:
807 cl_event ev = 0; 1739 EVENT_LIST (12);
1740
808 const size_t src_origin[3] = { src_x, src_y, src_z }; 1741 const size_t src_origin[3] = { src_x, src_y, src_z };
809 const size_t region[3] = { width, height, depth }; 1742 const size_t region[3] = { width, height, depth };
810 EVENT_LIST (12, items - 12);
811 1743
812 if (!row_pitch) 1744 if (!row_pitch)
813 row_pitch = img_row_pitch (src); 1745 row_pitch = img_row_pitch (src);
814 1746
815 if (depth > 1 && !slice_pitch) 1747 if (depth > 1 && !slice_pitch)
819 1751
820 SvUPGRADE (data, SVt_PV); 1752 SvUPGRADE (data, SVt_PV);
821 SvGROW (data, len); 1753 SvGROW (data, len);
822 SvPOK_only (data); 1754 SvPOK_only (data);
823 SvCUR_set (data, len); 1755 SvCUR_set (data, len);
1756
1757 cl_event ev = 0;
824 NEED_SUCCESS (EnqueueReadImage, (this, src, blocking, src_origin, region, row_pitch, slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1758 NEED_SUCCESS (EnqueueReadImage, (self, src, blocking, src_origin, region, row_pitch, slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
825 1759
826 if (ev) 1760 if (ev)
827 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1761 XPUSH_CLOBJ (stash_event, ev);
828 1762
829void 1763void
830enqueue_write_image (OpenCL::Queue this, OpenCL::Image dst, cl_bool blocking, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, size_t row_pitch, size_t slice_pitch, SV *data, ...) 1764write_image (OpenCL::Queue self, OpenCL::Image dst, cl_bool blocking, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, size_t row_pitch, size_t slice_pitch, SV *data, ...)
1765 ALIAS:
1766 enqueue_write_image = 0
831 PPCODE: 1767 PPCODE:
832 cl_event ev = 0; 1768 EVENT_LIST (12);
1769
833 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1770 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
834 const size_t region[3] = { width, height, depth }; 1771 const size_t region[3] = { width, height, depth };
835 STRLEN len; 1772 STRLEN len;
836 char *ptr = SvPVbyte (data, len); 1773 char *ptr = SvPVbyte (data, len);
837 EVENT_LIST (12, items - 12);
838 1774
839 if (!row_pitch) 1775 if (!row_pitch)
840 row_pitch = img_row_pitch (dst); 1776 row_pitch = img_row_pitch (dst);
841 1777
842 if (depth > 1 && !slice_pitch) 1778 if (depth > 1 && !slice_pitch)
845 size_t min_len = slice_pitch ? slice_pitch * depth : row_pitch * height; 1781 size_t min_len = slice_pitch ? slice_pitch * depth : row_pitch * height;
846 1782
847 if (len < min_len) 1783 if (len < min_len)
848 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred"); 1784 croak ("clEnqueueWriteImage: data string is shorter than what would be transferred");
849 1785
1786 cl_event ev = 0;
850 NEED_SUCCESS (EnqueueWriteImage, (this, dst, blocking, dst_origin, region, row_pitch, slice_pitch, SvPVX (data), event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1787 NEED_SUCCESS (EnqueueWriteImage, (self, dst, blocking, dst_origin, region, row_pitch, slice_pitch, ptr, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
851 1788
852 if (ev) 1789 if (ev)
853 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1790 XPUSH_CLOBJ (stash_event, ev);
854 1791
855void 1792void
856enqueue_copy_image (OpenCL::Queue this, OpenCL::Image src, OpenCL::Image dst, size_t src_x, size_t src_y, size_t src_z, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, ...) 1793copy_image (OpenCL::Queue self, OpenCL::Image src, OpenCL::Image dst, size_t src_x, size_t src_y, size_t src_z, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, ...)
1794 ALIAS:
1795 enqueue_copy_image = 0
857 PPCODE: 1796 PPCODE:
858 cl_event ev = 0; 1797 EVENT_LIST (12);
1798
859 const size_t src_origin[3] = { src_x, src_y, src_z }; 1799 const size_t src_origin[3] = { src_x, src_y, src_z };
860 const size_t dst_origin[3] = { dst_x, dst_y, dst_z }; 1800 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
861 const size_t region[3] = { width, height, depth }; 1801 const size_t region[3] = { width, height, depth };
862 EVENT_LIST (12, items - 12);
863 1802
1803 cl_event ev = 0;
864 NEED_SUCCESS (EnqueueCopyImage, (this, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1804 NEED_SUCCESS (EnqueueCopyImage, (self, src, dst, src_origin, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
865 1805
866 if (ev) 1806 if (ev)
867 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1807 XPUSH_CLOBJ (stash_event, ev);
868 1808
869void 1809void
870enqueue_copy_image_to_buffer (OpenCL::Queue this, OpenCL::Image src, OpenCL::Buffer dst, size_t src_x, size_t src_y, size_t src_z, size_t width, size_t height, size_t depth, size_t dst_offset, ...) 1810copy_image_to_buffer (OpenCL::Queue self, OpenCL::Image src, OpenCL::Buffer dst, size_t src_x, size_t src_y, size_t src_z, size_t width, size_t height, size_t depth, size_t dst_offset, ...)
1811 ALIAS:
1812 enqueue_copy_image_to_buffer = 0
871 PPCODE: 1813 PPCODE:
1814 EVENT_LIST (10);
1815
1816 const size_t src_origin[3] = { src_x, src_y, src_z };
1817 const size_t region [3] = { width, height, depth };
1818
872 cl_event ev = 0; 1819 cl_event ev = 0;
1820 NEED_SUCCESS (EnqueueCopyImageToBuffer, (self, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1821
1822 if (ev)
1823 XPUSH_CLOBJ (stash_event, ev);
1824
1825void
1826copy_buffer_to_image (OpenCL::Queue self, OpenCL::Buffer src, OpenCL::Image dst, size_t src_offset, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, ...)
1827 ALIAS:
1828 enqueue_copy_buffer_to_image = 0
1829 PPCODE:
1830 EVENT_LIST (10);
1831
873 const size_t src_origin[3] = { src_x, src_y, src_z }; 1832 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
1833 const size_t region [3] = { width, height, depth };
1834
1835 cl_event ev = 0;
1836 NEED_SUCCESS (EnqueueCopyBufferToImage, (self, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
1837
1838 if (ev)
1839 XPUSH_CLOBJ (stash_event, ev);
1840
1841void
1842map_buffer (OpenCL::Queue self, OpenCL::Buffer buf, cl_bool blocking = 1, cl_map_flags map_flags = CL_MAP_READ | CL_MAP_WRITE, size_t offset = 0, SV *cb_ = &PL_sv_undef, ...)
1843 ALIAS:
1844 enqueue_map_buffer = 0
1845 PPCODE:
1846 EVENT_LIST (6);
1847
1848 size_t cb = SvIV (cb_);
1849
1850 if (!SvOK (cb_))
1851 {
1852 NEED_SUCCESS (GetMemObjectInfo, (buf, CL_MEM_SIZE, sizeof (cb), &cb, 0));
1853 cb -= offset;
1854 }
1855
1856 cl_event ev;
1857 NEED_SUCCESS_ARG (void *ptr, EnqueueMapBuffer, (self, buf, blocking, map_flags, offset, cb, event_list_count, event_list_ptr, &ev, &res));
1858 XPUSHs (mapped_new (stash_mappedbuffer, self, buf, map_flags, ptr, cb, ev, 0, 0));
1859
1860void
1861map_image (OpenCL::Queue self, OpenCL::Image img, cl_bool blocking = 1, cl_map_flags map_flags = CL_MAP_READ | CL_MAP_WRITE, size_t x = 0, size_t y = 0, size_t z = 0, SV *width_ = &PL_sv_undef, SV *height_ = &PL_sv_undef, SV *depth_ = &PL_sv_undef, ...)
1862 ALIAS:
1863 enqueue_map_image = 0
1864 PPCODE:
1865 size_t width = SvIV (width_);
1866 if (!SvOK (width_))
1867 {
1868 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_WIDTH, sizeof (width), &width, 0));
1869 width -= x;
1870 }
1871
1872 size_t height = SvIV (width_);
1873 if (!SvOK (height_))
1874 {
1875 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_HEIGHT, sizeof (height), &height, 0));
1876 height -= y;
1877 }
1878
1879 size_t depth = SvIV (width_);
1880 if (!SvOK (depth_))
1881 {
1882 NEED_SUCCESS (GetImageInfo, (img, CL_IMAGE_DEPTH, sizeof (depth), &depth, 0));
1883 depth -= z;
1884
1885 // stupid opencl returns 0 for depth, but requires 1 for 2d images
1886 if (!depth)
1887 depth = 1;
1888 }
1889
1890 const size_t origin[3] = { x, y, z };
874 const size_t region[3] = { width, height, depth }; 1891 const size_t region[3] = { width, height, depth };
1892 size_t row_pitch, slice_pitch;
875 EVENT_LIST (10, items - 10); 1893 EVENT_LIST (10);
876 1894
877 NEED_SUCCESS (EnqueueCopyImageToBuffer, (this, src, dst, src_origin, region, dst_offset, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1895 cl_event ev;
1896 NEED_SUCCESS_ARG (void *ptr, EnqueueMapImage, (self, img, blocking, map_flags, origin, region, &row_pitch, &slice_pitch, event_list_count, event_list_ptr, &ev, &res));
1897
1898 size_t cb = slice_pitch ? slice_pitch * region [2]
1899 : row_pitch ? row_pitch * region [1]
1900 : region [0];
1901
1902 XPUSHs (mapped_new (stash_mappedimage, self, img, map_flags, ptr, cb, ev, row_pitch, slice_pitch));
1903
1904void
1905unmap (OpenCL::Queue self, OpenCL::Mapped mapped, ...)
1906 PPCODE:
1907 mapped_unmap (cv, ST (1), mapped, self, &ST (2), items - 2);
1908 if (GIMME_V != G_VOID)
1909 {
1910 clRetainEvent (mapped->event);
1911 XPUSH_CLOBJ (stash_event, mapped->event);
1912 }
1913
1914void
1915task (OpenCL::Queue self, OpenCL::Kernel kernel, ...)
1916 ALIAS:
1917 enqueue_task = 0
1918 PPCODE:
1919 EVENT_LIST (2);
1920
1921 cl_event ev = 0;
1922 NEED_SUCCESS (EnqueueTask, (self, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
878 1923
879 if (ev) 1924 if (ev)
880 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1925 XPUSH_CLOBJ (stash_event, ev);
881 1926
882void 1927void
883enqueue_copy_buffer_to_image (OpenCL::Queue this, OpenCL::Buffer src, OpenCL::Image dst, size_t src_offset, size_t dst_x, size_t dst_y, size_t dst_z, size_t width, size_t height, size_t depth, ...)
884 PPCODE:
885 cl_event ev = 0;
886 const size_t dst_origin[3] = { dst_x, dst_y, dst_z };
887 const size_t region[3] = { width, height, depth };
888 EVENT_LIST (10, items - 10);
889
890 NEED_SUCCESS (EnqueueCopyBufferToImage, (this, src, dst, src_offset, dst_origin, region, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
891
892 if (ev)
893 XPUSH_NEW_OBJ ("OpenCL::Event", ev);
894
895void
896enqueue_task (OpenCL::Queue this, OpenCL::Kernel kernel, ...)
897 PPCODE:
898 cl_event ev = 0;
899 EVENT_LIST (2, items - 2);
900
901 NEED_SUCCESS (EnqueueTask, (this, kernel, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
902
903 if (ev)
904 XPUSH_NEW_OBJ ("OpenCL::Event", ev);
905
906void
907enqueue_nd_range_kernel (OpenCL::Queue this, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...) 1928nd_range_kernel (OpenCL::Queue self, OpenCL::Kernel kernel, SV *global_work_offset, SV *global_work_size, SV *local_work_size = &PL_sv_undef, ...)
1929 ALIAS:
1930 enqueue_nd_range_kernel = 0
908 PPCODE: 1931 PPCODE:
909 cl_event ev = 0; 1932 EVENT_LIST (5);
1933
910 size_t *gwo = 0, *gws, *lws = 0; 1934 size_t *gwo = 0, *gws, *lws = 0;
911 int gws_len; 1935 int gws_len;
912 size_t *lists; 1936 size_t *lists;
913 int i; 1937 int i;
914 EVENT_LIST (5, items - 5);
915 1938
916 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV) 1939 if (!SvROK (global_work_size) || SvTYPE (SvRV (global_work_size)) != SVt_PVAV)
917 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference"); 1940 croak ("clEnqueueNDRangeKernel: global_work_size must be an array reference");
918 1941
919 gws_len = AvFILLp (SvRV (global_work_size)) + 1; 1942 gws_len = AvFILLp (SvRV (global_work_size)) + 1;
920 1943
921 lists = tmpbuf (sizeof (size_t) * 3 * gws_len); 1944 lists = tmpbuf (sizeof (size_t) * 3 * gws_len);
922 1945
923 gws = lists + gws_len * 0; 1946 gws = lists + gws_len * 0;
924 for (i = 0; i < gws_len; ++i) 1947 for (i = 0; i < gws_len; ++i)
1948 {
925 gws [i] = SvIV (AvARRAY (SvRV (global_work_size))[i]); 1949 gws [i] = SvIV (AvARRAY (SvRV (global_work_size))[i]);
1950 // at least nvidia crashes for 0-sized work group sizes, work around
1951 if (!gws [i])
1952 croak ("clEnqueueNDRangeKernel: global_work_size[%d] is zero, must be non-zero", i);
1953 }
926 1954
927 if (SvOK (global_work_offset)) 1955 if (SvOK (global_work_offset))
928 { 1956 {
929 if (!SvROK (global_work_offset) || SvTYPE (SvRV (global_work_offset)) != SVt_PVAV) 1957 if (!SvROK (global_work_offset) || SvTYPE (SvRV (global_work_offset)) != SVt_PVAV)
930 croak ("clEnqueueNDRangeKernel: global_work_offset must be undef or an array reference"); 1958 croak ("clEnqueueNDRangeKernel: global_work_offset must be undef or an array reference");
937 gwo [i] = SvIV (AvARRAY (SvRV (global_work_offset))[i]); 1965 gwo [i] = SvIV (AvARRAY (SvRV (global_work_offset))[i]);
938 } 1966 }
939 1967
940 if (SvOK (local_work_size)) 1968 if (SvOK (local_work_size))
941 { 1969 {
942 if (SvOK (local_work_size) && !SvROK (local_work_size) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV) 1970 if ((SvOK (local_work_size) && !SvROK (local_work_size)) || SvTYPE (SvRV (local_work_size)) != SVt_PVAV)
943 croak ("clEnqueueNDRangeKernel: global_work_size must be undef or an array reference"); 1971 croak ("clEnqueueNDRangeKernel: local_work_size must be undef or an array reference");
944 1972
945 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len) 1973 if (AvFILLp (SvRV (local_work_size)) + 1 != gws_len)
946 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size"); 1974 croak ("clEnqueueNDRangeKernel: local_work_local must be undef or an array of same size as global_work_size");
947 1975
948 lws = lists + gws_len * 2; 1976 lws = lists + gws_len * 2;
949 for (i = 0; i < gws_len; ++i) 1977 for (i = 0; i < gws_len; ++i)
1978 {
950 lws [i] = SvIV (AvARRAY (SvRV (local_work_size))[i]); 1979 lws [i] = SvIV (AvARRAY (SvRV (local_work_size))[i]);
1980 // at least nvidia crashes for 0-sized work group sizes, work around
1981 if (!lws [i])
1982 croak ("clEnqueueNDRangeKernel: local_work_size[%d] is zero, must be non-zero", i);
1983 }
951 } 1984 }
952 1985
1986 cl_event ev = 0;
953 NEED_SUCCESS (EnqueueNDRangeKernel, (this, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0)); 1987 NEED_SUCCESS (EnqueueNDRangeKernel, (self, kernel, gws_len, gwo, gws, lws, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
954 1988
955 if (ev) 1989 if (ev)
956 XPUSH_NEW_OBJ ("OpenCL::Event", ev); 1990 XPUSH_CLOBJ (stash_event, ev);
957 1991
1992#if CL_VERSION_1_2
1993
958void 1994void
959enqueue_marker (OpenCL::Queue this) 1995migrate_mem_objects (OpenCL::Queue self, SV *objects, cl_mem_migration_flags flags, ...)
1996 ALIAS:
1997 enqueue_migrate_mem_objects = 0
960 PPCODE: 1998 PPCODE:
1999 EVENT_LIST (3);
2000
2001 cl_uint obj_count;
2002 cl_mem *obj_list = object_list (cv, 0, "objects", objects, "OpenCL::Memory", &obj_count);
2003
2004 cl_event ev = 0;
2005 NEED_SUCCESS (EnqueueMigrateMemObjects, (self, obj_count, obj_list, flags, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2006
2007 if (ev)
2008 XPUSH_CLOBJ (stash_event, ev);
2009
2010#endif
2011
2012#if cl_apple_gl_sharing || cl_khr_gl_sharing
2013
2014void
2015acquire_gl_objects (OpenCL::Queue self, SV *objects, ...)
2016 ALIAS:
2017 release_gl_objects = 1
2018 enqueue_acquire_gl_objects = 0
2019 enqueue_release_gl_objects = 1
2020 PPCODE:
2021 EVENT_LIST (2);
2022
2023 cl_uint obj_count;
2024 cl_mem *obj_list = object_list (cv, 0, "objects", objects, "OpenCL::Memory", &obj_count);
2025
961 cl_event ev; 2026 cl_event ev = 0;
962 NEED_SUCCESS (EnqueueMarker, (this, &ev));
963 XPUSH_NEW_OBJ ("OpenCL::Event", ev);
964 2027
2028 if (ix)
2029 NEED_SUCCESS (EnqueueReleaseGLObjects, (self, obj_count, obj_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2030 else
2031 NEED_SUCCESS (EnqueueAcquireGLObjects, (self, obj_count, obj_list, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2032
2033 if (ev)
2034 XPUSH_CLOBJ (stash_event, ev);
2035
2036#endif
2037
965void 2038void
966enqueue_wait_for_events (OpenCL::Queue this, ...) 2039wait_for_events (OpenCL::Queue self, ...)
2040 ALIAS:
2041 enqueue_wait_for_events = 0
967 CODE: 2042 CODE:
968 EVENT_LIST (1, items - 1); 2043 EVENT_LIST (1);
2044#if PREFER_1_1
969 NEED_SUCCESS (EnqueueWaitForEvents, (this, event_list_count, event_list_ptr)); 2045 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
2046#else
2047 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, 0));
2048#endif
970 2049
971void 2050void
972enqueue_barrier (OpenCL::Queue this) 2051marker (OpenCL::Queue self, ...)
2052 ALIAS:
2053 enqueue_marker = 0
973 CODE: 2054 PPCODE:
2055 EVENT_LIST (1);
2056 cl_event ev = 0;
2057#if PREFER_1_1
2058 if (!event_list_count)
2059 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
2060 else
2061#if CL_VERSION_1_2
2062 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2063#else
2064 {
2065 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr)); // also a barrier
2066 NEED_SUCCESS (EnqueueMarker, (self, GIMME_V != G_VOID ? &ev : 0));
2067 }
2068#endif
2069#else
2070 NEED_SUCCESS (EnqueueMarkerWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2071#endif
2072 if (ev)
2073 XPUSH_CLOBJ (stash_event, ev);
2074
2075void
2076barrier (OpenCL::Queue self, ...)
2077 ALIAS:
2078 enqueue_barrier = 0
2079 PPCODE:
2080 EVENT_LIST (1);
2081 cl_event ev = 0;
2082#if PREFER_1_1
2083 if (!event_list_count && GIMME_V == G_VOID)
974 NEED_SUCCESS (EnqueueBarrier, (this)); 2084 NEED_SUCCESS (EnqueueBarrier, (self));
2085 else
2086#if CL_VERSION_1_2
2087 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2088#else
2089 {
2090 if (event_list_count)
2091 NEED_SUCCESS (EnqueueWaitForEvents, (self, event_list_count, event_list_ptr));
975 2092
2093 if (GIMME_V != G_VOID)
2094 NEED_SUCCESS (EnqueueMarker, (self, &ev));
2095 }
2096#endif
2097#else
2098 NEED_SUCCESS (EnqueueBarrierWithWaitList, (self, event_list_count, event_list_ptr, GIMME_V != G_VOID ? &ev : 0));
2099#endif
2100 if (ev)
2101 XPUSH_CLOBJ (stash_event, ev);
2102
976void 2103void
977flush (OpenCL::Queue this) 2104flush (OpenCL::Queue self)
978 CODE: 2105 CODE:
979 NEED_SUCCESS (Flush, (this)); 2106 NEED_SUCCESS (Flush, (self));
980 2107
981void 2108void
982finish (OpenCL::Queue this) 2109finish (OpenCL::Queue self)
983 CODE: 2110 CODE:
984 NEED_SUCCESS (Finish, (this)); 2111 NEED_SUCCESS (Finish, (self));
985 2112
986void 2113void
987info (OpenCL::Queue this, cl_command_queue_info name) 2114info (OpenCL::Queue self, cl_command_queue_info name)
988 PPCODE: 2115 PPCODE:
989 INFO (CommandQueue) 2116 INFO (CommandQueue)
990 2117
991#BEGIN:command_queue 2118#BEGIN:command_queue
992 2119
993void 2120void
994context (OpenCL::Queue this) 2121context (OpenCL::Queue self)
995 PPCODE: 2122 PPCODE:
996 cl_context value [1]; 2123 cl_context value [1];
997 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_CONTEXT, sizeof (value), value, 0)); 2124 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_CONTEXT, sizeof (value), value, 0));
998 EXTEND (SP, 1); 2125 EXTEND (SP, 1);
999 const int i = 0; 2126 const int i = 0;
1000 { 2127 {
1001 NEED_SUCCESS (RetainContext, (value [i])); 2128 NEED_SUCCESS (RetainContext, (value [i]));
1002 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2129 PUSH_CLOBJ (stash_context, value [i]);
1003 } 2130 }
1004 2131
1005void 2132void
1006device (OpenCL::Queue this) 2133device (OpenCL::Queue self)
1007 PPCODE: 2134 PPCODE:
1008 cl_device_id value [1]; 2135 cl_device_id value [1];
1009 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_DEVICE, sizeof (value), value, 0)); 2136 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_DEVICE, sizeof (value), value, 0));
1010 EXTEND (SP, 1); 2137 EXTEND (SP, 1);
1011 const int i = 0; 2138 const int i = 0;
1012 { 2139 {
1013 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 2140 PUSH_CLOBJ (stash_device, value [i]);
1014 } 2141 }
1015 2142
1016void 2143void
1017reference_count (OpenCL::Queue this) 2144reference_count (OpenCL::Queue self)
1018 PPCODE: 2145 PPCODE:
1019 cl_uint value [1]; 2146 cl_uint value [1];
1020 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_REFERENCE_COUNT, sizeof (value), value, 0)); 2147 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_REFERENCE_COUNT, sizeof (value), value, 0));
1021 EXTEND (SP, 1); 2148 EXTEND (SP, 1);
1022 const int i = 0; 2149 const int i = 0;
1023 PUSHs (sv_2mortal (newSVuv (value [i]))); 2150 PUSHs (sv_2mortal (newSVuv (value [i])));
1024 2151
1025void 2152void
1026properties (OpenCL::Queue this) 2153properties (OpenCL::Queue self)
1027 PPCODE: 2154 PPCODE:
1028 cl_command_queue_properties value [1]; 2155 cl_command_queue_properties value [1];
1029 NEED_SUCCESS (GetCommandQueueInfo, (this, CL_QUEUE_PROPERTIES, sizeof (value), value, 0)); 2156 NEED_SUCCESS (GetCommandQueueInfo, (self, CL_QUEUE_PROPERTIES, sizeof (value), value, 0));
1030 EXTEND (SP, 1); 2157 EXTEND (SP, 1);
1031 const int i = 0; 2158 const int i = 0;
1032 PUSHs (sv_2mortal (newSViv (value [i]))); 2159 PUSHs (sv_2mortal (newSVuv (value [i])));
1033 2160
1034#END:command_queue 2161#END:command_queue
1035 2162
1036MODULE = OpenCL PACKAGE = OpenCL::Memory 2163MODULE = OpenCL PACKAGE = OpenCL::Memory
1037 2164
1038void 2165void
1039DESTROY (OpenCL::Memory this) 2166DESTROY (OpenCL::Memory self)
1040 CODE: 2167 CODE:
1041 clReleaseMemObject (this); 2168 clReleaseMemObject (self);
1042 2169
1043void 2170void
1044info (OpenCL::Memory this, cl_mem_info name) 2171info (OpenCL::Memory self, cl_mem_info name)
1045 PPCODE: 2172 PPCODE:
1046 INFO (MemObject) 2173 INFO (MemObject)
1047 2174
1048#BEGIN:mem 2175#BEGIN:mem
1049 2176
1050void 2177void
1051type (OpenCL::Memory this) 2178type (OpenCL::Memory self)
1052 PPCODE: 2179 PPCODE:
1053 cl_mem_object_type value [1]; 2180 cl_mem_object_type value [1];
1054 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_TYPE, sizeof (value), value, 0)); 2181 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_TYPE, sizeof (value), value, 0));
1055 EXTEND (SP, 1); 2182 EXTEND (SP, 1);
1056 const int i = 0; 2183 const int i = 0;
1057 PUSHs (sv_2mortal (newSViv (value [i]))); 2184 PUSHs (sv_2mortal (newSVuv (value [i])));
1058 2185
1059void 2186void
1060flags (OpenCL::Memory this) 2187flags (OpenCL::Memory self)
1061 PPCODE: 2188 PPCODE:
1062 cl_mem_flags value [1]; 2189 cl_mem_flags value [1];
1063 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_FLAGS, sizeof (value), value, 0)); 2190 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_FLAGS, sizeof (value), value, 0));
1064 EXTEND (SP, 1); 2191 EXTEND (SP, 1);
1065 const int i = 0; 2192 const int i = 0;
1066 PUSHs (sv_2mortal (newSViv (value [i]))); 2193 PUSHs (sv_2mortal (newSVuv (value [i])));
1067 2194
1068void 2195void
1069size (OpenCL::Memory this) 2196size (OpenCL::Memory self)
1070 ALIAS: 2197 ALIAS:
1071 size = CL_MEM_SIZE 2198 size = CL_MEM_SIZE
1072 offset = CL_MEM_OFFSET 2199 offset = CL_MEM_OFFSET
1073 PPCODE: 2200 PPCODE:
1074 size_t value [1]; 2201 size_t value [1];
1075 NEED_SUCCESS (GetMemObjectInfo, (this, ix, sizeof (value), value, 0)); 2202 NEED_SUCCESS (GetMemObjectInfo, (self, ix, sizeof (value), value, 0));
1076 EXTEND (SP, 1); 2203 EXTEND (SP, 1);
1077 const int i = 0; 2204 const int i = 0;
1078 PUSHs (sv_2mortal (newSVuv (value [i]))); 2205 PUSHs (sv_2mortal (newSVuv (value [i])));
1079 2206
1080void 2207void
1081host_ptr (OpenCL::Memory this) 2208host_ptr (OpenCL::Memory self)
1082 PPCODE: 2209 PPCODE:
1083 void * value [1]; 2210 void * value [1];
1084 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_HOST_PTR, sizeof (value), value, 0)); 2211 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_HOST_PTR, sizeof (value), value, 0));
1085 EXTEND (SP, 1); 2212 EXTEND (SP, 1);
1086 const int i = 0; 2213 const int i = 0;
1087 PUSHs (sv_2mortal (newSVuv ((IV)(intptr_t)value [i]))); 2214 PUSHs (sv_2mortal (newSVuv ((IV)(intptr_t)value [i])));
1088 2215
1089void 2216void
1090map_count (OpenCL::Memory this) 2217map_count (OpenCL::Memory self)
1091 ALIAS: 2218 ALIAS:
1092 map_count = CL_MEM_MAP_COUNT 2219 map_count = CL_MEM_MAP_COUNT
1093 reference_count = CL_MEM_REFERENCE_COUNT 2220 reference_count = CL_MEM_REFERENCE_COUNT
1094 PPCODE: 2221 PPCODE:
1095 cl_uint value [1]; 2222 cl_uint value [1];
1096 NEED_SUCCESS (GetMemObjectInfo, (this, ix, sizeof (value), value, 0)); 2223 NEED_SUCCESS (GetMemObjectInfo, (self, ix, sizeof (value), value, 0));
1097 EXTEND (SP, 1); 2224 EXTEND (SP, 1);
1098 const int i = 0; 2225 const int i = 0;
1099 PUSHs (sv_2mortal (newSVuv (value [i]))); 2226 PUSHs (sv_2mortal (newSVuv (value [i])));
1100 2227
1101void 2228void
1102context (OpenCL::Memory this) 2229context (OpenCL::Memory self)
1103 PPCODE: 2230 PPCODE:
1104 cl_context value [1]; 2231 cl_context value [1];
1105 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_CONTEXT, sizeof (value), value, 0)); 2232 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_CONTEXT, sizeof (value), value, 0));
1106 EXTEND (SP, 1); 2233 EXTEND (SP, 1);
1107 const int i = 0; 2234 const int i = 0;
1108 { 2235 {
1109 NEED_SUCCESS (RetainContext, (value [i])); 2236 NEED_SUCCESS (RetainContext, (value [i]));
1110 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2237 PUSH_CLOBJ (stash_context, value [i]);
1111 } 2238 }
1112 2239
1113void 2240void
1114associated_memobject (OpenCL::Memory this) 2241associated_memobject (OpenCL::Memory self)
1115 PPCODE: 2242 PPCODE:
1116 cl_mem value [1]; 2243 cl_mem value [1];
1117 NEED_SUCCESS (GetMemObjectInfo, (this, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0)); 2244 NEED_SUCCESS (GetMemObjectInfo, (self, CL_MEM_ASSOCIATED_MEMOBJECT, sizeof (value), value, 0));
1118 EXTEND (SP, 1); 2245 EXTEND (SP, 1);
1119 const int i = 0; 2246 const int i = 0;
1120 { 2247 {
1121 NEED_SUCCESS (RetainMemObject, (value [i])); 2248 NEED_SUCCESS (RetainMemObject, (value [i]));
1122 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Memory", value [i])); 2249 PUSH_CLOBJ (stash_memory, value [i]);
1123 } 2250 }
1124 2251
1125#END:mem 2252#END:mem
1126 2253
2254#if cl_apple_gl_sharing || cl_khr_gl_sharing
2255
2256void
2257gl_object_info (OpenCL::Memory self)
2258 PPCODE:
2259 cl_gl_object_type type;
2260 cl_GLuint name;
2261 NEED_SUCCESS (GetGLObjectInfo, (self, &type, &name));
2262 EXTEND (SP, 2);
2263 PUSHs (sv_2mortal (newSVuv (type)));
2264 PUSHs (sv_2mortal (newSVuv (name)));
2265
2266#endif
2267
1127MODULE = OpenCL PACKAGE = OpenCL::BufferObj 2268MODULE = OpenCL PACKAGE = OpenCL::BufferObj
1128 2269
1129void 2270void
1130sub_buffer_region (OpenCL::BufferObj this, cl_mem_flags flags, size_t origin, size_t size) 2271sub_buffer_region (OpenCL::BufferObj self, cl_mem_flags flags, size_t origin, size_t size)
1131 PPCODE: 2272 PPCODE:
1132 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR | CL_MEM_ALLOC_HOST_PTR)) 2273 if (flags & (CL_MEM_USE_HOST_PTR | CL_MEM_COPY_HOST_PTR | CL_MEM_ALLOC_HOST_PTR))
1133 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!"); 2274 croak ("clCreateSubBuffer: cannot use/copy/alloc host ptr, doesn't make sense, check your flags!");
1134 2275
1135 cl_buffer_region crdata = { origin, size }; 2276 cl_buffer_region crdata = { origin, size };
1136 2277
1137 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (this, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res)); 2278 NEED_SUCCESS_ARG (cl_mem mem, CreateSubBuffer, (self, flags, CL_BUFFER_CREATE_TYPE_REGION, &crdata, &res));
1138 XPUSH_NEW_OBJ ("OpenCL::Buffer", mem); 2279 XPUSH_CLOBJ (stash_buffer, mem);
1139 2280
1140MODULE = OpenCL PACKAGE = OpenCL::Image 2281MODULE = OpenCL PACKAGE = OpenCL::Image
1141 2282
1142void 2283void
1143image_info (OpenCL::Image this, cl_image_info name) 2284image_info (OpenCL::Image self, cl_image_info name)
1144 PPCODE: 2285 PPCODE:
1145 INFO (Image) 2286 INFO (Image)
1146 2287
2288void
2289format (OpenCL::Image self)
2290 PPCODE:
2291 cl_image_format format;
2292 NEED_SUCCESS (GetImageInfo, (self, CL_IMAGE_FORMAT, sizeof (format), &format, 0));
2293 EXTEND (SP, 2);
2294 PUSHs (sv_2mortal (newSVuv (format.image_channel_order)));
2295 PUSHs (sv_2mortal (newSVuv (format.image_channel_data_type)));
2296
1147#BEGIN:image 2297#BEGIN:image
1148 2298
1149void 2299void
1150element_size (OpenCL::Image this) 2300element_size (OpenCL::Image self)
1151 ALIAS: 2301 ALIAS:
1152 element_size = CL_IMAGE_ELEMENT_SIZE 2302 element_size = CL_IMAGE_ELEMENT_SIZE
1153 row_pitch = CL_IMAGE_ROW_PITCH 2303 row_pitch = CL_IMAGE_ROW_PITCH
1154 slice_pitch = CL_IMAGE_SLICE_PITCH 2304 slice_pitch = CL_IMAGE_SLICE_PITCH
1155 width = CL_IMAGE_WIDTH 2305 width = CL_IMAGE_WIDTH
1156 height = CL_IMAGE_HEIGHT 2306 height = CL_IMAGE_HEIGHT
1157 depth = CL_IMAGE_DEPTH 2307 depth = CL_IMAGE_DEPTH
1158 PPCODE: 2308 PPCODE:
1159 size_t value [1]; 2309 size_t value [1];
1160 NEED_SUCCESS (GetImageInfo, (this, ix, sizeof (value), value, 0)); 2310 NEED_SUCCESS (GetImageInfo, (self, ix, sizeof (value), value, 0));
1161 EXTEND (SP, 1); 2311 EXTEND (SP, 1);
1162 const int i = 0; 2312 const int i = 0;
1163 PUSHs (sv_2mortal (newSVuv (value [i]))); 2313 PUSHs (sv_2mortal (newSVuv (value [i])));
1164 2314
1165#END:image 2315#END:image
1166 2316
2317#if cl_apple_gl_sharing || cl_khr_gl_sharing
2318
2319#BEGIN:gl_texture
2320
2321void
2322target (OpenCL::Image self)
2323 PPCODE:
2324 cl_GLenum value [1];
2325 NEED_SUCCESS (GetGLTextureInfo, (self, CL_GL_TEXTURE_TARGET, sizeof (value), value, 0));
2326 EXTEND (SP, 1);
2327 const int i = 0;
2328 PUSHs (sv_2mortal (newSVuv (value [i])));
2329
2330void
2331gl_mipmap_level (OpenCL::Image self)
2332 PPCODE:
2333 cl_GLint value [1];
2334 NEED_SUCCESS (GetGLTextureInfo, (self, CL_GL_MIPMAP_LEVEL, sizeof (value), value, 0));
2335 EXTEND (SP, 1);
2336 const int i = 0;
2337 PUSHs (sv_2mortal (newSViv (value [i])));
2338
2339#END:gl_texture
2340
2341#endif
2342
1167MODULE = OpenCL PACKAGE = OpenCL::Sampler 2343MODULE = OpenCL PACKAGE = OpenCL::Sampler
1168 2344
1169void 2345void
1170DESTROY (OpenCL::Sampler this) 2346DESTROY (OpenCL::Sampler self)
1171 CODE: 2347 CODE:
1172 clReleaseSampler (this); 2348 clReleaseSampler (self);
1173 2349
1174void 2350void
1175info (OpenCL::Sampler this, cl_sampler_info name) 2351info (OpenCL::Sampler self, cl_sampler_info name)
1176 PPCODE: 2352 PPCODE:
1177 INFO (Sampler) 2353 INFO (Sampler)
1178 2354
1179#BEGIN:sampler 2355#BEGIN:sampler
1180 2356
1181void 2357void
1182reference_count (OpenCL::Sampler this) 2358reference_count (OpenCL::Sampler self)
1183 PPCODE: 2359 PPCODE:
1184 cl_uint value [1]; 2360 cl_uint value [1];
1185 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_REFERENCE_COUNT, sizeof (value), value, 0)); 2361 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_REFERENCE_COUNT, sizeof (value), value, 0));
1186 EXTEND (SP, 1); 2362 EXTEND (SP, 1);
1187 const int i = 0; 2363 const int i = 0;
1188 PUSHs (sv_2mortal (newSVuv (value [i]))); 2364 PUSHs (sv_2mortal (newSVuv (value [i])));
1189 2365
1190void 2366void
1191context (OpenCL::Sampler this) 2367context (OpenCL::Sampler self)
1192 PPCODE: 2368 PPCODE:
1193 cl_context value [1]; 2369 cl_context value [1];
1194 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_CONTEXT, sizeof (value), value, 0)); 2370 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_CONTEXT, sizeof (value), value, 0));
1195 EXTEND (SP, 1); 2371 EXTEND (SP, 1);
1196 const int i = 0; 2372 const int i = 0;
1197 { 2373 {
1198 NEED_SUCCESS (RetainContext, (value [i])); 2374 NEED_SUCCESS (RetainContext, (value [i]));
1199 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2375 PUSH_CLOBJ (stash_context, value [i]);
1200 } 2376 }
1201 2377
1202void 2378void
1203normalized_coords (OpenCL::Sampler this) 2379normalized_coords (OpenCL::Sampler self)
1204 PPCODE: 2380 PPCODE:
1205 cl_addressing_mode value [1]; 2381 cl_addressing_mode value [1];
1206 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0)); 2382 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_NORMALIZED_COORDS, sizeof (value), value, 0));
1207 EXTEND (SP, 1); 2383 EXTEND (SP, 1);
1208 const int i = 0; 2384 const int i = 0;
1209 PUSHs (sv_2mortal (newSViv (value [i]))); 2385 PUSHs (sv_2mortal (newSVuv (value [i])));
1210 2386
1211void 2387void
1212addressing_mode (OpenCL::Sampler this) 2388addressing_mode (OpenCL::Sampler self)
1213 PPCODE: 2389 PPCODE:
1214 cl_filter_mode value [1]; 2390 cl_filter_mode value [1];
1215 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0)); 2391 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_ADDRESSING_MODE, sizeof (value), value, 0));
1216 EXTEND (SP, 1); 2392 EXTEND (SP, 1);
1217 const int i = 0; 2393 const int i = 0;
1218 PUSHs (sv_2mortal (newSViv (value [i]))); 2394 PUSHs (sv_2mortal (newSVuv (value [i])));
1219 2395
1220void 2396void
1221filter_mode (OpenCL::Sampler this) 2397filter_mode (OpenCL::Sampler self)
1222 PPCODE: 2398 PPCODE:
1223 cl_bool value [1]; 2399 cl_bool value [1];
1224 NEED_SUCCESS (GetSamplerInfo, (this, CL_SAMPLER_FILTER_MODE, sizeof (value), value, 0)); 2400 NEED_SUCCESS (GetSamplerInfo, (self, CL_SAMPLER_FILTER_MODE, sizeof (value), value, 0));
1225 EXTEND (SP, 1); 2401 EXTEND (SP, 1);
1226 const int i = 0; 2402 const int i = 0;
1227 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no)); 2403 PUSHs (sv_2mortal (value [i] ? &PL_sv_yes : &PL_sv_no));
1228 2404
1229#END:sampler 2405#END:sampler
1230 2406
1231MODULE = OpenCL PACKAGE = OpenCL::Program 2407MODULE = OpenCL PACKAGE = OpenCL::Program
1232 2408
1233void 2409void
1234DESTROY (OpenCL::Program this) 2410DESTROY (OpenCL::Program self)
1235 CODE: 2411 CODE:
1236 clReleaseProgram (this); 2412 clReleaseProgram (self);
1237 2413
1238void 2414void
1239build (OpenCL::Program this, OpenCL::Device device, SV *options = &PL_sv_undef) 2415build (OpenCL::Program self, SV *devices = &PL_sv_undef, SV *options = &PL_sv_undef, SV *notify = &PL_sv_undef)
2416 ALIAS:
2417 build_async = 1
1240 CODE: 2418 CODE:
2419 cl_uint device_count = 0;
2420 cl_device_id *device_list = 0;
2421
2422 if (SvOK (devices))
2423 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
2424
2425 void *user_data;
2426 program_callback pfn_notify = make_program_callback (notify, &user_data);
2427
2428 if (ix)
2429 build_program_async (self, device_count, device_list, SvPVbyte_nolen (options), user_data);
2430 else
2431 NEED_SUCCESS (BuildProgram, (self, device_count, device_list, SvPVbyte_nolen (options), pfn_notify, user_data));
2432
2433#if CL_VERSION_1_2
2434
2435void
2436compile (OpenCL::Program self, SV *devices, SV *options = &PL_sv_undef, SV *headers = &PL_sv_undef, SV *notify = &PL_sv_undef)
2437 CODE:
2438 cl_uint device_count = 0;
2439 cl_device_id *device_list = 0;
2440
2441 if (SvOK (devices))
2442 device_list = object_list (cv, 1, "devices", devices, "OpenCL::Device", &device_count);
2443
2444 cl_uint header_count = 0;
2445 cl_program *header_list = 0;
2446 const char **header_name = 0;
2447
2448 if (SvOK (headers))
2449 {
2450 if (!SvROK (devices) || SvTYPE (SvRV (devices)) != SVt_PVHV)
2451 croak ("clCompileProgram: headers must be undef or a hashref of name => OpenCL::Program pairs");
2452
2453 HV *hv = (HV *)SvRV (devices);
2454
2455 header_count = hv_iterinit (hv);
2456 header_list = tmpbuf (sizeof (*header_list) * header_count);
2457 header_name = tmpbuf (sizeof (*header_name) * header_count);
2458
2459 HE *he;
2460 int i = 0;
2461 while (he = hv_iternext (hv))
2462 {
2463 header_name [i] = SvPVbyte_nolen (HeSVKEY_force (he));
2464 header_list [i] = SvCLOBJ (cv, "headers", HeVAL (he), "OpenCL::Program");
2465 ++i;
2466 }
2467 }
2468
2469 void *user_data;
2470 program_callback pfn_notify = make_program_callback (notify, &user_data);
2471
1241 NEED_SUCCESS (BuildProgram, (this, 1, &device, SvPVbyte_nolen (options), 0, 0)); 2472 NEED_SUCCESS (CompileProgram, (self, device_count, device_list, SvPVbyte_nolen (options),
2473 header_count, header_list, header_name, pfn_notify, user_data));
1242 2474
2475#endif
2476
1243void 2477void
1244build_info (OpenCL::Program this, OpenCL::Device device, cl_program_build_info name) 2478build_info (OpenCL::Program self, OpenCL::Device device, cl_program_build_info name)
1245 PPCODE: 2479 PPCODE:
1246 size_t size; 2480 size_t size;
1247 NEED_SUCCESS (GetProgramBuildInfo, (this, device, name, 0, 0, &size)); 2481 NEED_SUCCESS (GetProgramBuildInfo, (self, device, name, 0, 0, &size));
1248 SV *sv = sv_2mortal (newSV (size)); 2482 SV *sv = sv_2mortal (newSV (size));
1249 SvUPGRADE (sv, SVt_PV); 2483 SvUPGRADE (sv, SVt_PV);
1250 SvPOK_only (sv); 2484 SvPOK_only (sv);
1251 SvCUR_set (sv, size); 2485 SvCUR_set (sv, size);
1252 NEED_SUCCESS (GetProgramBuildInfo, (this, device, name, size, SvPVX (sv), 0)); 2486 NEED_SUCCESS (GetProgramBuildInfo, (self, device, name, size, SvPVX (sv), 0));
1253 XPUSHs (sv); 2487 XPUSHs (sv);
1254 2488
1255#BEGIN:program_build 2489#BEGIN:program_build
1256 2490
1257void 2491void
1258build_status (OpenCL::Program this, OpenCL::Device device) 2492build_status (OpenCL::Program self, OpenCL::Device device)
1259 PPCODE: 2493 PPCODE:
1260 cl_build_status value [1]; 2494 cl_build_status value [1];
1261 NEED_SUCCESS (GetProgramBuildInfo, (this, device, CL_PROGRAM_BUILD_STATUS, sizeof (value), value, 0)); 2495 NEED_SUCCESS (GetProgramBuildInfo, (self, device, CL_PROGRAM_BUILD_STATUS, sizeof (value), value, 0));
1262 EXTEND (SP, 1); 2496 EXTEND (SP, 1);
1263 const int i = 0; 2497 const int i = 0;
1264 PUSHs (sv_2mortal (newSViv (value [i]))); 2498 PUSHs (sv_2mortal (newSViv (value [i])));
1265 2499
1266void 2500void
1267build_options (OpenCL::Program this, OpenCL::Device device) 2501build_options (OpenCL::Program self, OpenCL::Device device)
1268 ALIAS: 2502 ALIAS:
1269 build_options = CL_PROGRAM_BUILD_OPTIONS 2503 build_options = CL_PROGRAM_BUILD_OPTIONS
1270 build_log = CL_PROGRAM_BUILD_LOG 2504 build_log = CL_PROGRAM_BUILD_LOG
1271 PPCODE: 2505 PPCODE:
1272 size_t size; 2506 size_t size;
1273 NEED_SUCCESS (GetProgramBuildInfo, (this, device, ix, 0, 0, &size)); 2507 NEED_SUCCESS (GetProgramBuildInfo, (self, device, ix, 0, 0, &size));
1274 char *value = tmpbuf (size); 2508 char *value = tmpbuf (size);
1275 NEED_SUCCESS (GetProgramBuildInfo, (this, device, ix, size, value, 0)); 2509 NEED_SUCCESS (GetProgramBuildInfo, (self, device, ix, size, value, 0));
1276 EXTEND (SP, 1); 2510 EXTEND (SP, 1);
1277 const int i = 0; 2511 const int i = 0;
1278 PUSHs (sv_2mortal (newSVpv (value, 0))); 2512 PUSHs (sv_2mortal (newSVpv (value, 0)));
1279 2513
1280#END:program_build 2514#END:program_build
1281 2515
1282void 2516void
1283kernel (OpenCL::Program program, SV *function) 2517kernel (OpenCL::Program program, SV *function)
1284 PPCODE: 2518 PPCODE:
1285 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res)); 2519 NEED_SUCCESS_ARG (cl_kernel kernel, CreateKernel, (program, SvPVbyte_nolen (function), &res));
1286 XPUSH_NEW_OBJ ("OpenCL::Kernel", kernel); 2520 XPUSH_CLOBJ (stash_kernel, kernel);
1287 2521
1288void 2522void
2523kernels_in_program (OpenCL::Program program)
2524 PPCODE:
2525 cl_uint num_kernels;
2526 NEED_SUCCESS (CreateKernelsInProgram, (program, 0, 0, &num_kernels));
2527 cl_kernel *kernels = tmpbuf (sizeof (cl_kernel) * num_kernels);
2528 NEED_SUCCESS (CreateKernelsInProgram, (program, num_kernels, kernels, 0));
2529
2530 int i;
2531 EXTEND (SP, num_kernels);
2532 for (i = 0; i < num_kernels; ++i)
2533 PUSH_CLOBJ (stash_kernel, kernels [i]);
2534
2535void
1289info (OpenCL::Program this, cl_program_info name) 2536info (OpenCL::Program self, cl_program_info name)
1290 PPCODE: 2537 PPCODE:
1291 INFO (Program) 2538 INFO (Program)
1292 2539
1293void 2540void
1294binaries (OpenCL::Program this) 2541binaries (OpenCL::Program self)
1295 PPCODE: 2542 PPCODE:
1296 cl_uint n, i; 2543 cl_uint n, i;
1297 size_t size; 2544 size_t size;
1298 2545
1299 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_NUM_DEVICES , sizeof (n) , &n , 0)); 2546 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_NUM_DEVICES , sizeof (n) , &n , 0));
1300 if (!n) XSRETURN_EMPTY; 2547 if (!n) XSRETURN_EMPTY;
1301 2548
1302 size_t *sizes = tmpbuf (sizeof (*sizes) * n); 2549 size_t *sizes = tmpbuf (sizeof (*sizes) * n);
1303 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARY_SIZES, sizeof (*sizes) * n, sizes, &size)); 2550 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARY_SIZES, sizeof (*sizes) * n, sizes, &size));
1304 if (size != sizeof (*sizes) * n) XSRETURN_EMPTY; 2551 if (size != sizeof (*sizes) * n) XSRETURN_EMPTY;
1305 unsigned char **ptrs = tmpbuf (sizeof (*ptrs) * n); 2552 unsigned char **ptrs = tmpbuf (sizeof (*ptrs) * n);
1306 2553
1307 EXTEND (SP, n); 2554 EXTEND (SP, n);
1308 for (i = 0; i < n; ++i) 2555 for (i = 0; i < n; ++i)
1309 { 2556 {
1310 SV *sv = sv_2mortal (newSV (sizes [i])); 2557 SV *sv = sv_2mortal (newSV (sizes [i]));
1311 SvUPGRADE (sv, SVt_PV); 2558 SvUPGRADE (sv, SVt_PV);
1312 SvPOK_only (sv); 2559 SvPOK_only (sv);
1313 SvCUR_set (sv, sizes [i]); 2560 SvCUR_set (sv, sizes [i]);
1314 ptrs [i] = SvPVX (sv); 2561 ptrs [i] = (void *)SvPVX (sv);
1315 PUSHs (sv); 2562 PUSHs (sv);
1316 } 2563 }
1317 2564
1318 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARIES , sizeof (*ptrs ) * n, ptrs , &size)); 2565 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARIES , sizeof (*ptrs ) * n, ptrs , &size));
1319 if (size != sizeof (*ptrs) * n) XSRETURN_EMPTY; 2566 if (size != sizeof (*ptrs) * n) XSRETURN_EMPTY;
1320 2567
1321#BEGIN:program 2568#BEGIN:program
1322 2569
1323void 2570void
1324reference_count (OpenCL::Program this) 2571reference_count (OpenCL::Program self)
1325 ALIAS: 2572 ALIAS:
1326 reference_count = CL_PROGRAM_REFERENCE_COUNT 2573 reference_count = CL_PROGRAM_REFERENCE_COUNT
1327 num_devices = CL_PROGRAM_NUM_DEVICES 2574 num_devices = CL_PROGRAM_NUM_DEVICES
1328 PPCODE: 2575 PPCODE:
1329 cl_uint value [1]; 2576 cl_uint value [1];
1330 NEED_SUCCESS (GetProgramInfo, (this, ix, sizeof (value), value, 0)); 2577 NEED_SUCCESS (GetProgramInfo, (self, ix, sizeof (value), value, 0));
1331 EXTEND (SP, 1); 2578 EXTEND (SP, 1);
1332 const int i = 0; 2579 const int i = 0;
1333 PUSHs (sv_2mortal (newSVuv (value [i]))); 2580 PUSHs (sv_2mortal (newSVuv (value [i])));
1334 2581
1335void 2582void
1336context (OpenCL::Program this) 2583context (OpenCL::Program self)
1337 PPCODE: 2584 PPCODE:
1338 cl_context value [1]; 2585 cl_context value [1];
1339 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_CONTEXT, sizeof (value), value, 0)); 2586 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_CONTEXT, sizeof (value), value, 0));
1340 EXTEND (SP, 1); 2587 EXTEND (SP, 1);
1341 const int i = 0; 2588 const int i = 0;
1342 { 2589 {
1343 NEED_SUCCESS (RetainContext, (value [i])); 2590 NEED_SUCCESS (RetainContext, (value [i]));
1344 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2591 PUSH_CLOBJ (stash_context, value [i]);
1345 } 2592 }
1346 2593
1347void 2594void
1348devices (OpenCL::Program this) 2595devices (OpenCL::Program self)
1349 PPCODE: 2596 PPCODE:
1350 size_t size; 2597 size_t size;
1351 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_DEVICES, 0, 0, &size)); 2598 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, 0, 0, &size));
1352 cl_device_id *value = tmpbuf (size); 2599 cl_device_id *value = tmpbuf (size);
1353 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_DEVICES, size, value, 0)); 2600 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_DEVICES, size, value, 0));
1354 int i, n = size / sizeof (*value); 2601 int i, n = size / sizeof (*value);
1355 EXTEND (SP, n); 2602 EXTEND (SP, n);
1356 for (i = 0; i < n; ++i) 2603 for (i = 0; i < n; ++i)
1357 { 2604 {
1358 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Device", value [i])); 2605 PUSH_CLOBJ (stash_device, value [i]);
1359 } 2606 }
1360 2607
1361void 2608void
1362source (OpenCL::Program this) 2609source (OpenCL::Program self)
1363 PPCODE: 2610 PPCODE:
1364 size_t size; 2611 size_t size;
1365 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_SOURCE, 0, 0, &size)); 2612 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_SOURCE, 0, 0, &size));
1366 char *value = tmpbuf (size); 2613 char *value = tmpbuf (size);
1367 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_SOURCE, size, value, 0)); 2614 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_SOURCE, size, value, 0));
1368 EXTEND (SP, 1); 2615 EXTEND (SP, 1);
1369 const int i = 0; 2616 const int i = 0;
1370 PUSHs (sv_2mortal (newSVpv (value, 0))); 2617 PUSHs (sv_2mortal (newSVpv (value, 0)));
1371 2618
1372void 2619void
1373binary_sizes (OpenCL::Program this) 2620binary_sizes (OpenCL::Program self)
1374 PPCODE: 2621 PPCODE:
1375 size_t size; 2622 size_t size;
1376 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARY_SIZES, 0, 0, &size)); 2623 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARY_SIZES, 0, 0, &size));
1377 size_t *value = tmpbuf (size); 2624 size_t *value = tmpbuf (size);
1378 NEED_SUCCESS (GetProgramInfo, (this, CL_PROGRAM_BINARY_SIZES, size, value, 0)); 2625 NEED_SUCCESS (GetProgramInfo, (self, CL_PROGRAM_BINARY_SIZES, size, value, 0));
1379 int i, n = size / sizeof (*value); 2626 int i, n = size / sizeof (*value);
1380 EXTEND (SP, n); 2627 EXTEND (SP, n);
1381 for (i = 0; i < n; ++i) 2628 for (i = 0; i < n; ++i)
1382 PUSHs (sv_2mortal (newSVuv (value [i]))); 2629 PUSHs (sv_2mortal (newSVuv (value [i])));
1383 2630
1384#END:program 2631#END:program
1385 2632
1386MODULE = OpenCL PACKAGE = OpenCL::Kernel 2633MODULE = OpenCL PACKAGE = OpenCL::Kernel
1387 2634
1388void 2635void
1389DESTROY (OpenCL::Kernel this) 2636DESTROY (OpenCL::Kernel self)
1390 CODE: 2637 CODE:
1391 clReleaseKernel (this); 2638 clReleaseKernel (self);
1392 2639
1393void 2640void
2641setf (OpenCL::Kernel self, const char *format, ...)
2642 CODE:
2643 int i;
2644 for (i = 2; ; ++i)
2645 {
2646 while (*format == ' ')
2647 ++format;
2648
2649 char type = *format++;
2650
2651 if (!type)
2652 break;
2653
2654 if (i >= items)
2655 croak ("OpenCL::Kernel::setf format string too long (not enough arguments)");
2656
2657 SV *sv = ST (i);
2658
2659 union
2660 {
2661 cl_char cc; cl_uchar cC; cl_short cs; cl_ushort cS;
2662 cl_int ci; cl_uint cI; cl_long cl; cl_ulong cL;
2663 cl_half ch; cl_float cf; cl_double cd;
2664 cl_mem cm;
2665 cl_sampler ca;
2666 size_t cz;
2667 cl_event ce;
2668 } arg;
2669 size_t size;
2670 int nullarg = 0;
2671
2672 switch (type)
2673 {
2674 case 'c': arg.cc = SvIV (sv); size = sizeof (arg.cc); break;
2675 case 'C': arg.cC = SvUV (sv); size = sizeof (arg.cC); break;
2676 case 's': arg.cs = SvIV (sv); size = sizeof (arg.cs); break;
2677 case 'S': arg.cS = SvUV (sv); size = sizeof (arg.cS); break;
2678 case 'i': arg.ci = SvIV (sv); size = sizeof (arg.ci); break;
2679 case 'I': arg.cI = SvUV (sv); size = sizeof (arg.cI); break;
2680 case 'l': arg.cl = SvIV (sv); size = sizeof (arg.cl); break;
2681 case 'L': arg.cL = SvUV (sv); size = sizeof (arg.cL); break;
2682
2683 case 'h': arg.ch = SvUV (sv); size = sizeof (arg.ch); break;
2684 case 'f': arg.cf = SvNV (sv); size = sizeof (arg.cf); break;
2685 case 'd': arg.cd = SvNV (sv); size = sizeof (arg.cd); break;
2686
2687 case 'z': nullarg = 1; size = SvIV (sv); break;
2688
2689 case 'm': nullarg = !SvOK (sv); arg.cm = SvCLOBJ (cv, "m", sv, "OpenCL::Memory" ); size = sizeof (arg.cm); break;
2690 case 'a': nullarg = !SvOK (sv); arg.ca = SvCLOBJ (cv, "a", sv, "OpenCL::Sampler"); size = sizeof (arg.ca); break;
2691 case 'e': nullarg = !SvOK (sv); arg.ca = SvCLOBJ (cv, "e", sv, "OpenCL::Event" ); size = sizeof (arg.ce); break;
2692
2693 default:
2694 croak ("OpenCL::Kernel::setf format character '%c' not supported", type);
2695 }
2696
2697 res = clSetKernelArg (self, i - 2, size, nullarg ? 0 : &arg);
2698 if (res)
2699 croak ("OpenCL::Kernel::setf kernel parameter '%c' (#%d): %s", type, i - 2, err2str (res));
2700 }
2701
2702 if (i != items)
2703 croak ("OpenCL::Kernel::setf format string too short (too many arguments)");
2704
2705void
1394set_char (OpenCL::Kernel this, cl_uint idx, cl_char value) 2706set_char (OpenCL::Kernel self, cl_uint idx, cl_char value)
1395 CODE: 2707 CODE:
1396 clSetKernelArg (this, idx, sizeof (value), &value); 2708 clSetKernelArg (self, idx, sizeof (value), &value);
1397 2709
1398void 2710void
1399set_uchar (OpenCL::Kernel this, cl_uint idx, cl_uchar value) 2711set_uchar (OpenCL::Kernel self, cl_uint idx, cl_uchar value)
1400 CODE: 2712 CODE:
1401 clSetKernelArg (this, idx, sizeof (value), &value); 2713 clSetKernelArg (self, idx, sizeof (value), &value);
1402 2714
1403void 2715void
1404set_short (OpenCL::Kernel this, cl_uint idx, cl_short value) 2716set_short (OpenCL::Kernel self, cl_uint idx, cl_short value)
1405 CODE: 2717 CODE:
1406 clSetKernelArg (this, idx, sizeof (value), &value); 2718 clSetKernelArg (self, idx, sizeof (value), &value);
1407 2719
1408void 2720void
1409set_ushort (OpenCL::Kernel this, cl_uint idx, cl_ushort value) 2721set_ushort (OpenCL::Kernel self, cl_uint idx, cl_ushort value)
1410 CODE: 2722 CODE:
1411 clSetKernelArg (this, idx, sizeof (value), &value); 2723 clSetKernelArg (self, idx, sizeof (value), &value);
1412 2724
1413void 2725void
1414set_int (OpenCL::Kernel this, cl_uint idx, cl_int value) 2726set_int (OpenCL::Kernel self, cl_uint idx, cl_int value)
1415 CODE: 2727 CODE:
1416 clSetKernelArg (this, idx, sizeof (value), &value); 2728 clSetKernelArg (self, idx, sizeof (value), &value);
1417 2729
1418void 2730void
1419set_uint (OpenCL::Kernel this, cl_uint idx, cl_uint value) 2731set_uint (OpenCL::Kernel self, cl_uint idx, cl_uint value)
1420 CODE: 2732 CODE:
1421 clSetKernelArg (this, idx, sizeof (value), &value); 2733 clSetKernelArg (self, idx, sizeof (value), &value);
1422 2734
1423void 2735void
1424set_long (OpenCL::Kernel this, cl_uint idx, cl_long value) 2736set_long (OpenCL::Kernel self, cl_uint idx, cl_long value)
1425 CODE: 2737 CODE:
1426 clSetKernelArg (this, idx, sizeof (value), &value); 2738 clSetKernelArg (self, idx, sizeof (value), &value);
1427 2739
1428void 2740void
1429set_ulong (OpenCL::Kernel this, cl_uint idx, cl_ulong value) 2741set_ulong (OpenCL::Kernel self, cl_uint idx, cl_ulong value)
1430 CODE: 2742 CODE:
1431 clSetKernelArg (this, idx, sizeof (value), &value); 2743 clSetKernelArg (self, idx, sizeof (value), &value);
1432 2744
1433void 2745void
1434set_half (OpenCL::Kernel this, cl_uint idx, cl_half value) 2746set_half (OpenCL::Kernel self, cl_uint idx, cl_half value)
1435 CODE: 2747 CODE:
1436 clSetKernelArg (this, idx, sizeof (value), &value); 2748 clSetKernelArg (self, idx, sizeof (value), &value);
1437 2749
1438void 2750void
1439set_float (OpenCL::Kernel this, cl_uint idx, cl_float value) 2751set_float (OpenCL::Kernel self, cl_uint idx, cl_float value)
1440 CODE: 2752 CODE:
1441 clSetKernelArg (this, idx, sizeof (value), &value); 2753 clSetKernelArg (self, idx, sizeof (value), &value);
1442 2754
1443void 2755void
1444set_double (OpenCL::Kernel this, cl_uint idx, cl_double value) 2756set_double (OpenCL::Kernel self, cl_uint idx, cl_double value)
1445 CODE: 2757 CODE:
1446 clSetKernelArg (this, idx, sizeof (value), &value); 2758 clSetKernelArg (self, idx, sizeof (value), &value);
1447 2759
1448void 2760void
1449set_memory (OpenCL::Kernel this, cl_uint idx, OpenCL::Memory_ornull value) 2761set_memory (OpenCL::Kernel self, cl_uint idx, OpenCL::Memory_ornull value)
1450 CODE: 2762 CODE:
1451 clSetKernelArg (this, idx, sizeof (value), &value); 2763 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
1452 2764
1453void 2765void
1454set_buffer (OpenCL::Kernel this, cl_uint idx, OpenCL::Buffer_ornull value) 2766set_buffer (OpenCL::Kernel self, cl_uint idx, OpenCL::Buffer_ornull value)
1455 CODE: 2767 CODE:
1456 clSetKernelArg (this, idx, sizeof (value), &value); 2768 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
1457 2769
1458void 2770void
1459set_image2d (OpenCL::Kernel this, cl_uint idx, OpenCL::Image2D_ornull value) 2771set_image (OpenCL::Kernel self, cl_uint idx, OpenCL::Image_ornull value)
2772 ALIAS:
2773 set_image2d = 0
2774 set_image3d = 0
1460 CODE: 2775 CODE:
1461 clSetKernelArg (this, idx, sizeof (value), &value); 2776 clSetKernelArg (self, idx, sizeof (value), value ? &value : 0);
1462 2777
1463void 2778void
1464set_image3d (OpenCL::Kernel this, cl_uint idx, OpenCL::Image3D_ornull value)
1465 CODE:
1466 clSetKernelArg (this, idx, sizeof (value), &value);
1467
1468void
1469set_sampler (OpenCL::Kernel this, cl_uint idx, OpenCL::Sampler value) 2779set_sampler (OpenCL::Kernel self, cl_uint idx, OpenCL::Sampler value)
1470 CODE: 2780 CODE:
1471 clSetKernelArg (this, idx, sizeof (value), &value); 2781 clSetKernelArg (self, idx, sizeof (value), &value);
1472 2782
1473void 2783void
2784set_local (OpenCL::Kernel self, cl_uint idx, size_t size)
2785 CODE:
2786 clSetKernelArg (self, idx, size, 0);
2787
2788void
1474set_event (OpenCL::Kernel this, cl_uint idx, OpenCL::Event value) 2789set_event (OpenCL::Kernel self, cl_uint idx, OpenCL::Event value)
1475 CODE: 2790 CODE:
1476 clSetKernelArg (this, idx, sizeof (value), &value); 2791 clSetKernelArg (self, idx, sizeof (value), &value);
1477 2792
1478void 2793void
1479info (OpenCL::Kernel this, cl_kernel_info name) 2794info (OpenCL::Kernel self, cl_kernel_info name)
1480 PPCODE: 2795 PPCODE:
1481 INFO (Kernel) 2796 INFO (Kernel)
1482 2797
1483#BEGIN:kernel 2798#BEGIN:kernel
1484 2799
1485void 2800void
1486function_name (OpenCL::Kernel this) 2801function_name (OpenCL::Kernel self)
1487 PPCODE: 2802 PPCODE:
1488 size_t size; 2803 size_t size;
1489 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_FUNCTION_NAME, 0, 0, &size)); 2804 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_FUNCTION_NAME, 0, 0, &size));
1490 char *value = tmpbuf (size); 2805 char *value = tmpbuf (size);
1491 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_FUNCTION_NAME, size, value, 0)); 2806 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_FUNCTION_NAME, size, value, 0));
1492 EXTEND (SP, 1); 2807 EXTEND (SP, 1);
1493 const int i = 0; 2808 const int i = 0;
1494 PUSHs (sv_2mortal (newSVpv (value, 0))); 2809 PUSHs (sv_2mortal (newSVpv (value, 0)));
1495 2810
1496void 2811void
1497num_args (OpenCL::Kernel this) 2812num_args (OpenCL::Kernel self)
1498 ALIAS: 2813 ALIAS:
1499 num_args = CL_KERNEL_NUM_ARGS 2814 num_args = CL_KERNEL_NUM_ARGS
1500 reference_count = CL_KERNEL_REFERENCE_COUNT 2815 reference_count = CL_KERNEL_REFERENCE_COUNT
1501 PPCODE: 2816 PPCODE:
1502 cl_uint value [1]; 2817 cl_uint value [1];
1503 NEED_SUCCESS (GetKernelInfo, (this, ix, sizeof (value), value, 0)); 2818 NEED_SUCCESS (GetKernelInfo, (self, ix, sizeof (value), value, 0));
1504 EXTEND (SP, 1); 2819 EXTEND (SP, 1);
1505 const int i = 0; 2820 const int i = 0;
1506 PUSHs (sv_2mortal (newSVuv (value [i]))); 2821 PUSHs (sv_2mortal (newSVuv (value [i])));
1507 2822
1508void 2823void
1509context (OpenCL::Kernel this) 2824context (OpenCL::Kernel self)
1510 PPCODE: 2825 PPCODE:
1511 cl_context value [1]; 2826 cl_context value [1];
1512 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_CONTEXT, sizeof (value), value, 0)); 2827 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_CONTEXT, sizeof (value), value, 0));
1513 EXTEND (SP, 1); 2828 EXTEND (SP, 1);
1514 const int i = 0; 2829 const int i = 0;
1515 { 2830 {
1516 NEED_SUCCESS (RetainContext, (value [i])); 2831 NEED_SUCCESS (RetainContext, (value [i]));
1517 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 2832 PUSH_CLOBJ (stash_context, value [i]);
1518 } 2833 }
1519 2834
1520void 2835void
1521program (OpenCL::Kernel this) 2836program (OpenCL::Kernel self)
1522 PPCODE: 2837 PPCODE:
1523 cl_program value [1]; 2838 cl_program value [1];
1524 NEED_SUCCESS (GetKernelInfo, (this, CL_KERNEL_PROGRAM, sizeof (value), value, 0)); 2839 NEED_SUCCESS (GetKernelInfo, (self, CL_KERNEL_PROGRAM, sizeof (value), value, 0));
1525 EXTEND (SP, 1); 2840 EXTEND (SP, 1);
1526 const int i = 0; 2841 const int i = 0;
1527 { 2842 {
1528 NEED_SUCCESS (RetainProgram, (value [i])); 2843 NEED_SUCCESS (RetainProgram, (value [i]));
1529 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Program", value [i])); 2844 PUSH_CLOBJ (stash_program, value [i]);
1530 } 2845 }
1531 2846
1532#END:kernel 2847#END:kernel
1533 2848
1534void 2849void
1535work_group_info (OpenCL::Kernel this, OpenCL::Device device, cl_kernel_work_group_info name) 2850work_group_info (OpenCL::Kernel self, OpenCL::Device device, cl_kernel_work_group_info name)
1536 PPCODE: 2851 PPCODE:
1537 size_t size; 2852 size_t size;
1538 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, name, 0, 0, &size)); 2853 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, name, 0, 0, &size));
1539 SV *sv = sv_2mortal (newSV (size)); 2854 SV *sv = sv_2mortal (newSV (size));
1540 SvUPGRADE (sv, SVt_PV); 2855 SvUPGRADE (sv, SVt_PV);
1541 SvPOK_only (sv); 2856 SvPOK_only (sv);
1542 SvCUR_set (sv, size); 2857 SvCUR_set (sv, size);
1543 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, name, size, SvPVX (sv), 0)); 2858 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, name, size, SvPVX (sv), 0));
1544 XPUSHs (sv); 2859 XPUSHs (sv);
1545 2860
1546#BEGIN:kernel_work_group 2861#BEGIN:kernel_work_group
1547 2862
1548void 2863void
1549work_group_size (OpenCL::Kernel this, OpenCL::Device device) 2864work_group_size (OpenCL::Kernel self, OpenCL::Device device)
1550 ALIAS: 2865 ALIAS:
1551 work_group_size = CL_KERNEL_WORK_GROUP_SIZE 2866 work_group_size = CL_KERNEL_WORK_GROUP_SIZE
1552 preferred_work_group_size_multiple = CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE 2867 preferred_work_group_size_multiple = CL_KERNEL_PREFERRED_WORK_GROUP_SIZE_MULTIPLE
1553 PPCODE: 2868 PPCODE:
1554 size_t value [1]; 2869 size_t value [1];
1555 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, ix, sizeof (value), value, 0)); 2870 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, ix, sizeof (value), value, 0));
1556 EXTEND (SP, 1); 2871 EXTEND (SP, 1);
1557 const int i = 0; 2872 const int i = 0;
1558 PUSHs (sv_2mortal (newSVuv (value [i]))); 2873 PUSHs (sv_2mortal (newSVuv (value [i])));
1559 2874
1560void 2875void
1561compile_work_group_size (OpenCL::Kernel this, OpenCL::Device device) 2876compile_work_group_size (OpenCL::Kernel self, OpenCL::Device device)
1562 PPCODE: 2877 PPCODE:
1563 size_t size; 2878 size_t size;
1564 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, 0, 0, &size)); 2879 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, 0, 0, &size));
1565 size_t *value = tmpbuf (size); 2880 size_t *value = tmpbuf (size);
1566 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, size, value, 0)); 2881 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, CL_KERNEL_COMPILE_WORK_GROUP_SIZE, size, value, 0));
1567 int i, n = size / sizeof (*value); 2882 int i, n = size / sizeof (*value);
1568 EXTEND (SP, n); 2883 EXTEND (SP, n);
1569 for (i = 0; i < n; ++i) 2884 for (i = 0; i < n; ++i)
1570 PUSHs (sv_2mortal (newSVuv (value [i]))); 2885 PUSHs (sv_2mortal (newSVuv (value [i])));
1571 2886
1572void 2887void
1573local_mem_size (OpenCL::Kernel this, OpenCL::Device device) 2888local_mem_size (OpenCL::Kernel self, OpenCL::Device device)
1574 ALIAS: 2889 ALIAS:
1575 local_mem_size = CL_KERNEL_LOCAL_MEM_SIZE 2890 local_mem_size = CL_KERNEL_LOCAL_MEM_SIZE
1576 private_mem_size = CL_KERNEL_PRIVATE_MEM_SIZE 2891 private_mem_size = CL_KERNEL_PRIVATE_MEM_SIZE
1577 PPCODE: 2892 PPCODE:
1578 cl_ulong value [1]; 2893 cl_ulong value [1];
1579 NEED_SUCCESS (GetKernelWorkGroupInfo, (this, device, ix, sizeof (value), value, 0)); 2894 NEED_SUCCESS (GetKernelWorkGroupInfo, (self, device, ix, sizeof (value), value, 0));
1580 EXTEND (SP, 1); 2895 EXTEND (SP, 1);
1581 const int i = 0; 2896 const int i = 0;
1582 PUSHs (sv_2mortal (newSVuv (value [i]))); 2897 PUSHs (sv_2mortal (newSVuv (value [i])));
1583 2898
1584#END:kernel_work_group 2899#END:kernel_work_group
1585 2900
2901#if CL_VERSION_1_2
2902
2903void
2904arg_info (OpenCL::Kernel self, cl_uint idx, cl_kernel_arg_info name)
2905 PPCODE:
2906 size_t size;
2907 NEED_SUCCESS (GetKernelArgInfo, (self, idx, name, 0, 0, &size));
2908 SV *sv = sv_2mortal (newSV (size));
2909 SvUPGRADE (sv, SVt_PV);
2910 SvPOK_only (sv);
2911 SvCUR_set (sv, size);
2912 NEED_SUCCESS (GetKernelArgInfo, (self, idx, name, size, SvPVX (sv), 0));
2913 XPUSHs (sv);
2914
2915#BEGIN:kernel_arg
2916
2917void
2918arg_address_qualifier (OpenCL::Kernel self, cl_uint idx)
2919 PPCODE:
2920 cl_kernel_arg_address_qualifier value [1];
2921 NEED_SUCCESS (GetKernelArgInfo, (self, idx, CL_KERNEL_ARG_ADDRESS_QUALIFIER, sizeof (value), value, 0));
2922 EXTEND (SP, 1);
2923 const int i = 0;
2924 PUSHs (sv_2mortal (newSVuv (value [i])));
2925
2926void
2927arg_access_qualifier (OpenCL::Kernel self, cl_uint idx)
2928 PPCODE:
2929 cl_kernel_arg_access_qualifier value [1];
2930 NEED_SUCCESS (GetKernelArgInfo, (self, idx, CL_KERNEL_ARG_ACCESS_QUALIFIER, sizeof (value), value, 0));
2931 EXTEND (SP, 1);
2932 const int i = 0;
2933 PUSHs (sv_2mortal (newSVuv (value [i])));
2934
2935void
2936arg_type_name (OpenCL::Kernel self, cl_uint idx)
2937 ALIAS:
2938 arg_type_name = CL_KERNEL_ARG_TYPE_NAME
2939 arg_name = CL_KERNEL_ARG_NAME
2940 PPCODE:
2941 size_t size;
2942 NEED_SUCCESS (GetKernelArgInfo, (self, idx, ix, 0, 0, &size));
2943 char *value = tmpbuf (size);
2944 NEED_SUCCESS (GetKernelArgInfo, (self, idx, ix, size, value, 0));
2945 EXTEND (SP, 1);
2946 const int i = 0;
2947 PUSHs (sv_2mortal (newSVpv (value, 0)));
2948
2949void
2950arg_type_qualifier (OpenCL::Kernel self, cl_uint idx)
2951 PPCODE:
2952 cl_kernel_arg_type_qualifier value [1];
2953 NEED_SUCCESS (GetKernelArgInfo, (self, idx, CL_KERNEL_ARG_TYPE_QUALIFIER, sizeof (value), value, 0));
2954 EXTEND (SP, 1);
2955 const int i = 0;
2956 PUSHs (sv_2mortal (newSVuv (value [i])));
2957
2958#END:kernel_arg
2959
2960#endif
2961
1586MODULE = OpenCL PACKAGE = OpenCL::Event 2962MODULE = OpenCL PACKAGE = OpenCL::Event
1587 2963
1588void 2964void
1589DESTROY (OpenCL::Event this) 2965DESTROY (OpenCL::Event self)
1590 CODE: 2966 CODE:
1591 clReleaseEvent (this); 2967 clReleaseEvent (self);
1592 2968
1593void 2969void
1594wait (OpenCL::Event this) 2970wait (OpenCL::Event self)
1595 CODE: 2971 CODE:
1596 clWaitForEvents (1, &this); 2972 clWaitForEvents (1, &self);
1597 2973
1598void 2974void
2975cb (OpenCL::Event self, cl_int command_exec_callback_type, SV *cb)
2976 CODE:
2977 clSetEventCallback (self, command_exec_callback_type, eq_event_notify, SvREFCNT_inc (s_get_cv (cb)));
2978
2979void
1599info (OpenCL::Event this, cl_event_info name) 2980info (OpenCL::Event self, cl_event_info name)
1600 PPCODE: 2981 PPCODE:
1601 INFO (Event) 2982 INFO (Event)
1602 2983
1603#BEGIN:event 2984#BEGIN:event
1604 2985
1605void 2986void
1606command_queue (OpenCL::Event this) 2987command_queue (OpenCL::Event self)
1607 PPCODE: 2988 PPCODE:
1608 cl_command_queue value [1]; 2989 cl_command_queue value [1];
1609 NEED_SUCCESS (GetEventInfo, (this, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0)); 2990 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_QUEUE, sizeof (value), value, 0));
1610 EXTEND (SP, 1); 2991 EXTEND (SP, 1);
1611 const int i = 0; 2992 const int i = 0;
1612 { 2993 {
1613 NEED_SUCCESS (RetainCommandQueue, (value [i])); 2994 NEED_SUCCESS (RetainCommandQueue, (value [i]));
1614 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Queue", value [i])); 2995 PUSH_CLOBJ (stash_queue, value [i]);
1615 } 2996 }
1616 2997
1617void 2998void
1618command_type (OpenCL::Event this) 2999command_type (OpenCL::Event self)
1619 PPCODE: 3000 PPCODE:
1620 cl_command_type value [1]; 3001 cl_command_type value [1];
1621 NEED_SUCCESS (GetEventInfo, (this, CL_EVENT_COMMAND_TYPE, sizeof (value), value, 0)); 3002 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_COMMAND_TYPE, sizeof (value), value, 0));
1622 EXTEND (SP, 1); 3003 EXTEND (SP, 1);
1623 const int i = 0; 3004 const int i = 0;
1624 PUSHs (sv_2mortal (newSVuv (value [i]))); 3005 PUSHs (sv_2mortal (newSVuv (value [i])));
1625 3006
1626void 3007void
1627reference_count (OpenCL::Event this) 3008reference_count (OpenCL::Event self)
1628 ALIAS: 3009 ALIAS:
1629 reference_count = CL_EVENT_REFERENCE_COUNT 3010 reference_count = CL_EVENT_REFERENCE_COUNT
1630 command_execution_status = CL_EVENT_COMMAND_EXECUTION_STATUS 3011 command_execution_status = CL_EVENT_COMMAND_EXECUTION_STATUS
1631 PPCODE: 3012 PPCODE:
1632 cl_uint value [1]; 3013 cl_uint value [1];
1633 NEED_SUCCESS (GetEventInfo, (this, ix, sizeof (value), value, 0)); 3014 NEED_SUCCESS (GetEventInfo, (self, ix, sizeof (value), value, 0));
1634 EXTEND (SP, 1); 3015 EXTEND (SP, 1);
1635 const int i = 0; 3016 const int i = 0;
1636 PUSHs (sv_2mortal (newSVuv (value [i]))); 3017 PUSHs (sv_2mortal (newSVuv (value [i])));
1637 3018
1638void 3019void
1639context (OpenCL::Event this) 3020context (OpenCL::Event self)
1640 PPCODE: 3021 PPCODE:
1641 cl_context value [1]; 3022 cl_context value [1];
1642 NEED_SUCCESS (GetEventInfo, (this, CL_EVENT_CONTEXT, sizeof (value), value, 0)); 3023 NEED_SUCCESS (GetEventInfo, (self, CL_EVENT_CONTEXT, sizeof (value), value, 0));
1643 EXTEND (SP, 1); 3024 EXTEND (SP, 1);
1644 const int i = 0; 3025 const int i = 0;
1645 { 3026 {
1646 NEED_SUCCESS (RetainContext, (value [i])); 3027 NEED_SUCCESS (RetainContext, (value [i]));
1647 PUSHs (NEW_MORTAL_OBJ ("OpenCL::Context", value [i])); 3028 PUSH_CLOBJ (stash_context, value [i]);
1648 } 3029 }
1649 3030
1650#END:event 3031#END:event
1651 3032
1652void 3033void
1653profiling_info (OpenCL::Event this, cl_profiling_info name) 3034profiling_info (OpenCL::Event self, cl_profiling_info name)
1654 PPCODE: 3035 PPCODE:
1655 INFO (EventProfiling) 3036 INFO (EventProfiling)
1656 3037
1657#BEGIN:profiling 3038#BEGIN:profiling
1658 3039
1659void 3040void
1660profiling_command_queued (OpenCL::Event this) 3041profiling_command_queued (OpenCL::Event self)
1661 ALIAS: 3042 ALIAS:
1662 profiling_command_queued = CL_PROFILING_COMMAND_QUEUED 3043 profiling_command_queued = CL_PROFILING_COMMAND_QUEUED
1663 profiling_command_submit = CL_PROFILING_COMMAND_SUBMIT 3044 profiling_command_submit = CL_PROFILING_COMMAND_SUBMIT
1664 profiling_command_start = CL_PROFILING_COMMAND_START 3045 profiling_command_start = CL_PROFILING_COMMAND_START
1665 profiling_command_end = CL_PROFILING_COMMAND_END 3046 profiling_command_end = CL_PROFILING_COMMAND_END
1666 PPCODE: 3047 PPCODE:
1667 cl_ulong value [1]; 3048 cl_ulong value [1];
1668 NEED_SUCCESS (GetEventProfilingInfo, (this, ix, sizeof (value), value, 0)); 3049 NEED_SUCCESS (GetEventProfilingInfo, (self, ix, sizeof (value), value, 0));
1669 EXTEND (SP, 1); 3050 EXTEND (SP, 1);
1670 const int i = 0; 3051 const int i = 0;
1671 PUSHs (sv_2mortal (newSVuv (value [i]))); 3052 PUSHs (sv_2mortal (newSVuv (value [i])));
1672 3053
1673#END:profiling 3054#END:profiling
1674 3055
1675MODULE = OpenCL PACKAGE = OpenCL::UserEvent 3056MODULE = OpenCL PACKAGE = OpenCL::UserEvent
1676 3057
1677void 3058void
1678set_status (OpenCL::UserEvent this, cl_int execution_status) 3059set_status (OpenCL::UserEvent self, cl_int execution_status)
1679 CODE: 3060 CODE:
1680 clSetUserEventStatus (this, execution_status); 3061 clSetUserEventStatus (self, execution_status);
1681 3062
3063MODULE = OpenCL PACKAGE = OpenCL::Mapped
3064
3065void
3066DESTROY (SV *self)
3067 CODE:
3068 OpenCL__Mapped mapped = SvMAPPED (self);
3069
3070 clEnqueueUnmapMemObject (mapped->queue, mapped->memobj, mapped->ptr, 1, &mapped->event, 0);
3071 mapped_detach (self, mapped);
3072
3073 clReleaseCommandQueue (mapped->queue);
3074 clReleaseEvent (mapped->event);
3075 Safefree (mapped);
3076
3077void
3078unmap (OpenCL::Mapped self, ...)
3079 CODE:
3080 mapped_unmap (cv, ST (0), self, self->queue, &ST (1), items - 1);
3081
3082bool
3083mapped (OpenCL::Mapped self)
3084 CODE:
3085 RETVAL = !!self->ptr;
3086 OUTPUT:
3087 RETVAL
3088
3089void
3090wait (OpenCL::Mapped self)
3091 PPCODE:
3092 if (self->event)
3093 NEED_SUCCESS (WaitForEvents, (1, &self->event));
3094
3095void
3096event (OpenCL::Mapped self)
3097 PPCODE:
3098 if (!self->event)
3099 XSRETURN_UNDEF;
3100
3101 clRetainEvent (self->event);
3102 XPUSH_CLOBJ (stash_event, self->event);
3103
3104size_t
3105size (OpenCL::Mapped self)
3106 CODE:
3107 RETVAL = self->cb;
3108 OUTPUT:
3109 RETVAL
3110
3111IV
3112ptr (OpenCL::Mapped self)
3113 CODE:
3114 RETVAL = PTR2IV (self->ptr);
3115 OUTPUT:
3116 RETVAL
3117
3118void
3119set (OpenCL::Mapped self, size_t offset, SV *data)
3120 CODE:
3121 STRLEN len;
3122 const char *ptr = SvPVbyte (data, len);
3123
3124 if (offset + len > self->cb)
3125 croak ("OpenCL::Mapped::set out of bound condition detected");
3126
3127 memcpy (offset + (char *)self->ptr, ptr, len);
3128
3129MODULE = OpenCL PACKAGE = OpenCL::MappedBuffer
3130
3131MODULE = OpenCL PACKAGE = OpenCL::MappedImage
3132
3133IV
3134row_pitch (OpenCL::Mapped self)
3135 ALIAS:
3136 slice_pitch = 1
3137 CODE:
3138 RETVAL = ix ? self->slice_pitch : self->row_pitch;
3139 OUTPUT:
3140 RETVAL
3141
3142
3143

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines